From marty at toombs.com Fri May 1 01:02:29 2009 From: marty at toombs.com (Marty Toombs) Date: Thu, 30 Apr 2009 20:02:29 -0400 Subject: Whither Novell In-Reply-To: References: Message-ID: <49FA3C15.2000109@toombs.com> I have a slightly different take on the Netware issue and what versions to compare to what. I'd like to compare Netware 7 to Windows Server 2008. But I can't do that. I feel that Netware has left me and the fact that Novell is no longer producing new versions was their choice, not mine. I'm not leaving Novell, they left me. I'm the IS director for a small non-profit. In today's environment I can (and have) argued for Netware for all the reasons everyone on this list knows. At this point, those reasons don't exist for Linux. At the same time I'm dealing with more and more applications which are Windows only. Our payroll program used to run on Netware, now it's Windows only, etc. I've been forced to set up Windows servers next to my Netware boxes just to keep doing what I'm doing. As I look to the future, I think it would be irresponsible for me to go to SUSE servers when I can't be certain that proper support for them will be available. I have to think of the agency's future and not just my wishes. So the end of Netware for me has left me with just one choice -- how fast do we move to Windows. My current plan calls for the last Netware server to come down some time next year. I will not enjoy that when it happens. From petervl at gmail.com Fri May 1 03:34:17 2009 From: petervl at gmail.com (Peter Van Lone) Date: Thu, 30 Apr 2009 21:34:17 -0500 Subject: NW6.0 to NW6.5 in-place upgrade - "rollback" option? In-Reply-To: <49F9D571020000750004C47F@inet.eastcobbgroup.com> References: <68b791330904300507s1198ff5as8353fa4d8016ae84@mail.gmail.com> <20090430121246.2CCBB242F5@webmail218.herald.ox.ac.uk> <68b791330904300629p6e1e6cddkcb9daf767ac89bb9@mail.gmail.com> <49F97FF9020000750003152C@inet.eastcobbgroup.com> <68b791330904300858vcc49563s331f8ce8b9f3fe8d@mail.gmail.com> <49F9E9D7.6050902@calfrye.com> <68b791330904301227k71f72e2ch8de8c3bd92c2c9e4@mail.gmail.com> <49F9D571020000750004C47F@inet.eastcobbgroup.com> Message-ID: <68b791330904301934q788d6516p8a7fde1137a4bb7c@mail.gmail.com> On Thu, Apr 30, 2009 at 3:44 PM, James Taylor wrote: > Keep in mind that a new install of NW6.5SP7 will install eDir 8.8, no options. > The only way to keep eDir 8.7.9 is to upgrade from a prior release. > > The biggest issue with eDir 8.8 in an old environment would be if you have really old, or unpatched versions of ZEN or BorderManager. > Otherwise, 8.8 should not be a problem. > > I haven't run into a site where it has been an issue after we did a complete configuration and prerequisites review. it is precisely ZFD 7 (unpatched) that is on the server in question, and very likely on other servers in the tree. And it is used extensively and relied upon, even as it is being backed away from and systematically replaced by SMS (or whatever they call it, now). I am tasked with helping them keep things stable and working, while the rug is changed out to MS sigh I suppose I could try to find an SP6 over-lay CD to install from, that way it would not introduce the 8.8 issues. The latest version they have on any server is sp7, and it is not at edir 8.8 P From randygrein at comcast.net Fri May 1 04:02:56 2009 From: randygrein at comcast.net (Randy Grein) Date: Thu, 30 Apr 2009 20:02:56 -0700 Subject: Whither Novell In-Reply-To: <49FA3C15.2000109@toombs.com> References: <49FA3C15.2000109@toombs.com> Message-ID: Sorry Marty, but I can't buy the logic you're using. Can you explain more? Linux has steadily increased in market share, eating mostly into Netware (but, AFAICT Windows as well) so that it's somewhere between 25-35% of new installs. Support is available from multiple vendors (open source, SUSE, Red Hat, Debian, Ubuntu, etc) so what's the real problem? Keep in mind the distinction between security and profit. Security calls for following the herd. Profit calls for striking out on your own (correct) path. How much do you trust your own analysis? Randy Grein, Master CNE, CCNA On Apr 30, 2009, at 5:02 PM, Marty Toombs wrote: > I have a slightly different take on the Netware issue and what > versions to compare to what. > > I'd like to compare Netware 7 to Windows Server 2008. But I can't do > that. > > I feel that Netware has left me and the fact that Novell is no > longer producing new versions was their choice, not mine. I'm not > leaving Novell, they left me. > > I'm the IS director for a small non-profit. In today's environment I > can (and have) argued for Netware for all the reasons everyone on > this list knows. At this point, those reasons don't exist for Linux. > At the same time I'm dealing with more and more applications which > are Windows only. Our payroll program used to run on Netware, now > it's Windows only, etc. I've been forced to set up Windows servers > next to my Netware boxes just to keep doing what I'm doing. > > As I look to the future, I think it would be irresponsible for me to > go to SUSE servers when I can't be certain that proper support for > them will be available. I have to think of the agency's future and > not just my wishes. > > So the end of Netware for me has left me with just one choice -- how > fast do we move to Windows. > > My current plan calls for the last Netware server to come down some > time next year. I will not enjoy that when it happens. > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From cmangiarelli at gmail.com Fri May 1 06:28:00 2009 From: cmangiarelli at gmail.com (Christopher Mangiarelli) Date: Fri, 1 May 2009 01:28:00 -0400 Subject: NW6.0 to NW6.5 in-place upgrade - "rollback" option? In-Reply-To: <68b791330904301934q788d6516p8a7fde1137a4bb7c@mail.gmail.com> References: <68b791330904300507s1198ff5as8353fa4d8016ae84@mail.gmail.com> <20090430121246.2CCBB242F5@webmail218.herald.ox.ac.uk> <68b791330904300629p6e1e6cddkcb9daf767ac89bb9@mail.gmail.com> <49F97FF9020000750003152C@inet.eastcobbgroup.com> <68b791330904300858vcc49563s331f8ce8b9f3fe8d@mail.gmail.com> <49F9E9D7.6050902@calfrye.com> <68b791330904301227k71f72e2ch8de8c3bd92c2c9e4@mail.gmail.com> <49F9D571020000750004C47F@inet.eastcobbgroup.com> <68b791330904301934q788d6516p8a7fde1137a4bb7c@mail.gmail.com> Message-ID: <393865BA-0F21-4F5D-B5AB-DAF43D170774@gmail.com> I've done several in place upgrades from nw6 to nw65. Used a sp6 overlay cd. Like joe said, make sure you have space and Edir is healthy. Obviously backups are important. Get the environment to edir 8.7.3.10b and nw65sp7 after all is well. Can also go to sp8 too. Both are working well for me. Sent from my iPhone On Apr 30, 2009, at 10:34 PM, Peter Van Lone wrote: > On Thu, Apr 30, 2009 at 3:44 PM, James Taylor > wrote: >> Keep in mind that a new install of NW6.5SP7 will install eDir 8.8, >> no options. >> The only way to keep eDir 8.7.9 is to upgrade from a prior release. >> >> The biggest issue with eDir 8.8 in an old environment would be if >> you have really old, or unpatched versions of ZEN or BorderManager. >> Otherwise, 8.8 should not be a problem. >> >> I haven't run into a site where it has been an issue after we did a >> complete configuration and prerequisites review. > > > it is precisely ZFD 7 (unpatched) that is on the server in question, > and very likely on other servers in the tree. And it is used > extensively and relied upon, even as it is being backed away from and > systematically replaced by SMS (or whatever they call it, now). > > I am tasked with helping them keep things stable and working, while > the rug is changed out to MS > > sigh > > I suppose I could try to find an SP6 over-lay CD to install from, that > way it would not introduce the 8.8 issues. The latest version they > have on any server is sp7, and it is not at edir 8.8 > > P > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From petervl at gmail.com Fri May 1 12:04:53 2009 From: petervl at gmail.com (Peter Van Lone) Date: Fri, 1 May 2009 06:04:53 -0500 Subject: NW6.0 to NW6.5 in-place upgrade - "rollback" option? In-Reply-To: <393865BA-0F21-4F5D-B5AB-DAF43D170774@gmail.com> References: <68b791330904300507s1198ff5as8353fa4d8016ae84@mail.gmail.com> <20090430121246.2CCBB242F5@webmail218.herald.ox.ac.uk> <68b791330904300629p6e1e6cddkcb9daf767ac89bb9@mail.gmail.com> <49F97FF9020000750003152C@inet.eastcobbgroup.com> <68b791330904300858vcc49563s331f8ce8b9f3fe8d@mail.gmail.com> <49F9E9D7.6050902@calfrye.com> <68b791330904301227k71f72e2ch8de8c3bd92c2c9e4@mail.gmail.com> <49F9D571020000750004C47F@inet.eastcobbgroup.com> <68b791330904301934q788d6516p8a7fde1137a4bb7c@mail.gmail.com> <393865BA-0F21-4F5D-B5AB-DAF43D170774@gmail.com> Message-ID: <68b791330905010404ja526ea7x5c5e65822e9f26fc@mail.gmail.com> On Fri, May 1, 2009 at 12:28 AM, Christopher Mangiarelli wrote: > I've done several in place upgrades from nw6 to nw65. ?Used a sp6 overlay > cd. which method did you use? "Down server" upgrade where you boot to the ne 65 overlay CD? > Like joe said, make sure you have space and Edir is healthy. Obviously > backups are important. Get the environment to edir 8.7.3.10b and nw65sp7 > after all is well. Can also go to sp8 too. Both are working well for me. thnx! From mrsmith at oconee.k12.ga.us Fri May 1 13:23:16 2009 From: mrsmith at oconee.k12.ga.us (Matt Smith) Date: Fri, 01 May 2009 08:23:16 -0400 Subject: Whither Novell In-Reply-To: References: <49FA3C15.2000109@toombs.com> Message-ID: <49FAB16B.E4C1.0068.0@oconee.k12.ga.us> >>> On 4/30/2009 at 11:02 PM, in message , Randy Grein wrote: > Sorry Marty, but I can't buy the logic you're using. Can you explain > more? > > Linux has steadily increased in market share, eating mostly into > Netware (but, AFAICT Windows as well) so that it's somewhere between > 25-35% of new installs. Support is available from multiple vendors > (open source, SUSE, Red Hat, Debian, Ubuntu, etc) so what's the real > problem? > > Keep in mind the distinction between security and profit. Security > calls for following the herd. Profit calls for striking out on your > own (correct) path. How much do you trust your own analysis? I don't want to "gang up" on you, Marty, but my first response was pretty much the same as Randy's here. We do have several windows servers running as application servers. However, I've found lately that we were installing Windows to run these application not because they were Windows only, but because the sales people pitched it as a Windows application to the power-that-be. Turns out that if we'd asked, we could have been running some of this stuff on Linux after all. I'm particularly burned by the recent purchase of our HR system. The company that owns the product actually develops on Linux and ports it over to Windows! I'm not being ingenuous. Windows dominates the market. Still, my boss loves to brag when he hears the latest school system that lost a day or two's work to the latest internet nasty while we soldier on without a hitch. Besides, SUSE is linux. I figure if I have to, we'll just switch to a different flavor. I don't expect my experience to be quite as rich as what I get with Novell, but I also don't feel like I'm wasting my time now by any means. Each to their own though. -Matt -- Matt Smith Network Technology Specialist Oconee County School System, Oconee County, Georgia Office of Instruction and Technology 706-769-5685 x1314 From joea at j4computers.com Fri May 1 15:58:43 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 01 May 2009 10:58:43 -0400 Subject: 2 node cluster Message-ID: <49FAD5D3020000850005FC1E@FS-LIN-OES> Testing shows working as expected. Pull plug on slave node, it takes a dive, reboots, rejoins. Same for Master, except resources move over to slave, and (former) slave assumes Master IP. What happens in the event of a switch power loss, or network crash, seems a bit problematic. I have not been brave enough, or cruel enough (poor little servers) to pull the plug on both, at the same instant. From a Human perspective. We do, if we know there is going to be network stuff going on, break the clusters to avoid any issues. joe a. >>> "joea at j4computers.com" 04/30/09 11:45 AM >>> My understanding is (was) that there are several levels of decision making before biting down hard on the molar. The TID refers to OES2, we're on OES1 and will be for a while. Not to imply a similar bug cannot exist. Hmm, dawns on me there is a test node I can just go pull the plugs on, in deliberate fashion, and find out. Eh, eh? See, still working. Hope I remember to write down the results. joe a. >>> "John Cunningham" 04/30/09 11:01 AM >>> Hm. Would the non-master node, as determined by the SBD partition, not recognize that it is not a master, cannot get a heartbeat packet from the master and promptly commit Harakiri? I was under the impression that the SBD partition was the ultimate authority when it came to node voting. Otherwise, I would think it'd be a bug in the SBD code. Not that THAT's ever happened... *cough* http://www.novell.com/support/viewContent.do?externalId=3767762&sliceId=1 *cough* I agree though: 2 nodes, a cluster does not make. ~John >>> On 4/30/2009 at 11:31 AM, Jon Christensen wrote: That's the problem with two node clusters. Under certain conditions (rare, but possible) you can end up with a split brain condition, with both nodes thinking they are the master & both writing to the shared storage. This will result in data corruption. Many people will not make a cluster smaller than three nodes because of this. I've installed quite a few two node clusters & haven't seen problems, but the possibility does exist. My caveat - I haven't done much with clustering for the past two years, so things may have changed. Jon On Thu, Apr 30, 2009 at 8:20 AM, joea at j4computers.com wrote: > Actually, it was reading that document that spurred my question. > > The doc seems to be saying that, in a split brain condition, two masters > will exist, hence the attempt to determine LAN connectivity. But, if the > comms issue was fleeting, and now comms are "good", what then? We have two > masters, good comms, who wins? Does Cluster Services retain knowledge of > who was the previous Master, and it wins? > > The root of this is an infrequent (rare, actually) reboot of non master > nodes. OES1, Linux, 2 node cluster. > > joe a. > > >>> "John Cunningham" 04/30/09 9:47 AM >>> > Split brain condition? > > I keep referring back to TID 10053882 when it comes to this stuff. > > http://support.novell.com/docs/Tids/Solutions/10053882.html > > >>> On 4/30/2009 at 10:42 AM, Bill Brush wrote: > I think you have it right if my sessions at Brainshare are coming back > to me correctly. > > Bill > > On Thu, Apr 30, 2009 at 8:01 AM, joea at j4computers.com > wrote: > > Somebody know cluster services? > > > > In a 2 node cluster, docs seem to say a special condition exists, (when > there is a loss of heartbeat) in that, since there cannot be a "majority", > nor can the Master node be assumed "good", then, if network connectivity can > be determined, the "bad" node eats a poison pill. > > > > However, docs do not state what happens when connectivity *cannot* be > determined. Does the Master then always win? From experience, it would seem > so, as we have had instances of the non master node rebooting, apparently > spontaneously. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- My blog: http://www.antioxidantalley.com/blog LinkedIn Profile: http://www.linkedin.com/in/jbchristensen Facebook Profile: http://profile.to/jbchristensen/ Follow me on Twitter: http://twitter.com/jbchristensen _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From Rpd at co.mason.wa.us Fri May 1 16:55:01 2009 From: Rpd at co.mason.wa.us (Bob Deans) Date: Fri, 01 May 2009 08:55:01 -0700 Subject: 2 node cluster In-Reply-To: <49FAD5D3020000850005FC1E@FS-LIN-OES> References: <49FAD5D3020000850005FC1E@FS-LIN-OES> Message-ID: <49FAB8E3.6CD7.0020.1@co.mason.wa.us> I have been using two node clusters for six years. I am using HP's Cluster In A Box. I have had numerous power outages over the years and had both nodes go down at the same time and have never had any problems. Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> "joea at j4computers.com" 5/1/2009 7:58 AM >>> Testing shows working as expected. Pull plug on slave node, it takes a dive, reboots, rejoins. Same for Master, except resources move over to slave, and (former) slave assumes Master IP. What happens in the event of a switch power loss, or network crash, seems a bit problematic. I have not been brave enough, or cruel enough (poor little servers) to pull the plug on both, at the same instant. From a Human perspective. We do, if we know there is going to be network stuff going on, break the clusters to avoid any issues. joe a. >>> "joea at j4computers.com" 04/30/09 11:45 AM >>> My understanding is (was) that there are several levels of decision making before biting down hard on the molar. The TID refers to OES2, we're on OES1 and will be for a while. Not to imply a similar bug cannot exist. Hmm, dawns on me there is a test node I can just go pull the plugs on, in deliberate fashion, and find out. Eh, eh? See, still working. Hope I remember to write down the results. joe a. >>> "John Cunningham" 04/30/09 11:01 AM >>> Hm. Would the non-master node, as determined by the SBD partition, not recognize that it is not a master, cannot get a heartbeat packet from the master and promptly commit Harakiri? I was under the impression that the SBD partition was the ultimate authority when it came to node voting. Otherwise, I would think it'd be a bug in the SBD code. Not that THAT's ever happened... *cough* http://www.novell.com/support/viewContent.do?externalId=3767762&sliceId=1 *cough* I agree though: 2 nodes, a cluster does not make. ~John >>> On 4/30/2009 at 11:31 AM, Jon Christensen wrote: That's the problem with two node clusters. Under certain conditions (rare, but possible) you can end up with a split brain condition, with both nodes thinking they are the master & both writing to the shared storage. This will result in data corruption. Many people will not make a cluster smaller than three nodes because of this. I've installed quite a few two node clusters & haven't seen problems, but the possibility does exist. My caveat - I haven't done much with clustering for the past two years, so things may have changed. Jon On Thu, Apr 30, 2009 at 8:20 AM, joea at j4computers.com wrote: > Actually, it was reading that document that spurred my question. > > The doc seems to be saying that, in a split brain condition, two masters > will exist, hence the attempt to determine LAN connectivity. But, if the > comms issue was fleeting, and now comms are "good", what then? We have two > masters, good comms, who wins? Does Cluster Services retain knowledge of > who was the previous Master, and it wins? > > The root of this is an infrequent (rare, actually) reboot of non master > nodes. OES1, Linux, 2 node cluster. > > joe a. > > >>> "John Cunningham" 04/30/09 9:47 AM >>> > Split brain condition? > > I keep referring back to TID 10053882 when it comes to this stuff. > > http://support.novell.com/docs/Tids/Solutions/10053882.html > > >>> On 4/30/2009 at 10:42 AM, Bill Brush wrote: > I think you have it right if my sessions at Brainshare are coming back > to me correctly. > > Bill > > On Thu, Apr 30, 2009 at 8:01 AM, joea at j4computers.com > wrote: > > Somebody know cluster services? > > > > In a 2 node cluster, docs seem to say a special condition exists, (when > there is a loss of heartbeat) in that, since there cannot be a "majority", > nor can the Master node be assumed "good", then, if network connectivity can > be determined, the "bad" node eats a poison pill. > > > > However, docs do not state what happens when connectivity *cannot* be > determined. Does the Master then always win? From experience, it would seem > so, as we have had instances of the non master node rebooting, apparently > spontaneously. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- My blog: http://www.antioxidantalley.com/blog LinkedIn Profile: http://www.linkedin.com/in/jbchristensen Facebook Profile: http://profile.to/jbchristensen/ Follow me on Twitter: http://twitter.com/jbchristensen _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Fri May 1 18:56:51 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 01 May 2009 13:56:51 -0400 Subject: OES1 used swap space growing, does not return. Message-ID: <49FAFF9B020000850005FC28@FS-LIN-OES> GW7.0.3 hp2. OES1 Linux. 2 node cluster. 3-5 POA's per server. Swap space in use is increasing dramatically on certain servers after having been stable for a long time. The only change was to increase the frequency of Quickfinder indexing. Changed from once a day, to once every 4 hours. Staggered the start times, on any given box, so the jobs did not overlap each other, based on the maximum time QF was observed to run, looking at POA logs. This did not seem too insane, based on reading various blurbs. Seems each job only runs a few minutes. Yet seems QF is eating a lot of memory, causing swap to increase, and it never seems to come back to any great degree. I have adjusted the start and interval, run only twice a day, yet the swap space in use continues to grow, causing some concern. I have added swap files, with same priorty as existing. Swap part and orginal swap are now almost "full" after having dropped immediately after adding swap and added swap is beginning to be used. What can I look for? joe a. From joea at j4computers.com Fri May 1 19:44:09 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 01 May 2009 14:44:09 -0400 Subject: 2 node cluster Message-ID: <49FB0AA7020000850005FC35@FS-LIN-OES> An interesting point of fact. But unless that is an OEM version of NCS, your experience is not necessarily applicable. Likely HP's own concoction with different philosophy, design, coding. joe a. >>> "Bob Deans" 05/01/09 11:56 AM >>> I have been using two node clusters for six years. I am using HP's Cluster In A Box. I have had numerous power outages over the years and had both nodes go down at the same time and have never had any problems. Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> "joea at j4computers.com" 5/1/2009 7:58 AM >>> Testing shows working as expected. Pull plug on slave node, it takes a dive, reboots, rejoins. Same for Master, except resources move over to slave, and (former) slave assumes Master IP. What happens in the event of a switch power loss, or network crash, seems a bit problematic. I have not been brave enough, or cruel enough (poor little servers) to pull the plug on both, at the same instant. From a Human perspective. We do, if we know there is going to be network stuff going on, break the clusters to avoid any issues. joe a. >>> "joea at j4computers.com" 04/30/09 11:45 AM >>> My understanding is (was) that there are several levels of decision making before biting down hard on the molar. The TID refers to OES2, we're on OES1 and will be for a while. Not to imply a similar bug cannot exist. Hmm, dawns on me there is a test node I can just go pull the plugs on, in deliberate fashion, and find out. Eh, eh? See, still working. Hope I remember to write down the results. joe a. >>> "John Cunningham" 04/30/09 11:01 AM >>> Hm. Would the non-master node, as determined by the SBD partition, not recognize that it is not a master, cannot get a heartbeat packet from the master and promptly commit Harakiri? I was under the impression that the SBD partition was the ultimate authority when it came to node voting. Otherwise, I would think it'd be a bug in the SBD code. Not that THAT's ever happened... *cough* http://www.novell.com/support/viewContent.do?externalId=3767762&sliceId=1 *cough* I agree though: 2 nodes, a cluster does not make. ~John >>> On 4/30/2009 at 11:31 AM, Jon Christensen wrote: That's the problem with two node clusters. Under certain conditions (rare, but possible) you can end up with a split brain condition, with both nodes thinking they are the master & both writing to the shared storage. This will result in data corruption. Many people will not make a cluster smaller than three nodes because of this. I've installed quite a few two node clusters & haven't seen problems, but the possibility does exist. My caveat - I haven't done much with clustering for the past two years, so things may have changed. Jon On Thu, Apr 30, 2009 at 8:20 AM, joea at j4computers.com wrote: > Actually, it was reading that document that spurred my question. > > The doc seems to be saying that, in a split brain condition, two masters > will exist, hence the attempt to determine LAN connectivity. But, if the > comms issue was fleeting, and now comms are "good", what then? We have two > masters, good comms, who wins? Does Cluster Services retain knowledge of > who was the previous Master, and it wins? > > The root of this is an infrequent (rare, actually) reboot of non master > nodes. OES1, Linux, 2 node cluster. > > joe a. > > >>> "John Cunningham" 04/30/09 9:47 AM >>> > Split brain condition? > > I keep referring back to TID 10053882 when it comes to this stuff. > > http://support.novell.com/docs/Tids/Solutions/10053882.html > > >>> On 4/30/2009 at 10:42 AM, Bill Brush wrote: > I think you have it right if my sessions at Brainshare are coming back > to me correctly. > > Bill > > On Thu, Apr 30, 2009 at 8:01 AM, joea at j4computers.com > wrote: > > Somebody know cluster services? > > > > In a 2 node cluster, docs seem to say a special condition exists, (when > there is a loss of heartbeat) in that, since there cannot be a "majority", > nor can the Master node be assumed "good", then, if network connectivity can > be determined, the "bad" node eats a poison pill. > > > > However, docs do not state what happens when connectivity *cannot* be > determined. Does the Master then always win? From experience, it would seem > so, as we have had instances of the non master node rebooting, apparently > spontaneously. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- My blog: http://www.antioxidantalley.com/blog LinkedIn Profile: http://www.linkedin.com/in/jbchristensen Facebook Profile: http://profile.to/jbchristensen/ Follow me on Twitter: http://twitter.com/jbchristensen _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From RGrein at tpchd.org Fri May 1 20:01:34 2009 From: RGrein at tpchd.org (Randy Grein) Date: Fri, 01 May 2009 12:01:34 -0700 Subject: 2 node cluster In-Reply-To: <49FB0AA7020000850005FC35@FS-LIN-OES> References: <49FB0AA7020000850005FC35@FS-LIN-OES> Message-ID: <49FAE49E0200007200032B85@health-mail2.tpchd.org> No, I've worked with HP's offering and it's straight Netware. That is, the hardware is from HP, uses (or used) a shared SCSI connect to storage and a pair of their servers all in a single enclosure. If memory serves it comes with a NIC for heartbeat connection between the servers but that may have been an earlier model - Bob would know what he has and can correct my details. In any case, the likelihood is that if not there's an included ethernet switch that would provide connections. A lot of ways to slice this puppy. (Now THERE's an image!) Randy Grein Sr. Network Engineer >>> "joea at j4computers.com" 5/1/2009 11:44 AM >>> An interesting point of fact. But unless that is an OEM version of NCS, your experience is not necessarily applicable. Likely HP's own concoction with different philosophy, design, coding. joe a. >>> "Bob Deans" 05/01/09 11:56 AM >>> I have been using two node clusters for six years. I am using HP's Cluster In A Box. I have had numerous power outages over the years and had both nodes go down at the same time and have never had any problems. Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> "joea at j4computers.com" 5/1/2009 7:58 AM >>> Testing shows working as expected. Pull plug on slave node, it takes a dive, reboots, rejoins. Same for Master, except resources move over to slave, and (former) slave assumes Master IP. What happens in the event of a switch power loss, or network crash, seems a bit problematic. I have not been brave enough, or cruel enough (poor little servers) to pull the plug on both, at the same instant. From a Human perspective. We do, if we know there is going to be network stuff going on, break the clusters to avoid any issues. joe a. >>> "joea at j4computers.com" 04/30/09 11:45 AM >>> My understanding is (was) that there are several levels of decision making before biting down hard on the molar. The TID refers to OES2, we're on OES1 and will be for a while. Not to imply a similar bug cannot exist. Hmm, dawns on me there is a test node I can just go pull the plugs on, in deliberate fashion, and find out. Eh, eh? See, still working. Hope I remember to write down the results. joe a. >>> "John Cunningham" 04/30/09 11:01 AM >>> Hm. Would the non-master node, as determined by the SBD partition, not recognize that it is not a master, cannot get a heartbeat packet from the master and promptly commit Harakiri? I was under the impression that the SBD partition was the ultimate authority when it came to node voting. Otherwise, I would think it'd be a bug in the SBD code. Not that THAT's ever happened... *cough* http://www.novell.com/support/viewContent.do?externalId=3767762&sliceId=1 *cough* I agree though: 2 nodes, a cluster does not make. ~John >>> On 4/30/2009 at 11:31 AM, Jon Christensen wrote: That's the problem with two node clusters. Under certain conditions (rare, but possible) you can end up with a split brain condition, with both nodes thinking they are the master & both writing to the shared storage. This will result in data corruption. Many people will not make a cluster smaller than three nodes because of this. I've installed quite a few two node clusters & haven't seen problems, but the possibility does exist. My caveat - I haven't done much with clustering for the past two years, so things may have changed. Jon On Thu, Apr 30, 2009 at 8:20 AM, joea at j4computers.com wrote: > Actually, it was reading that document that spurred my question. > > The doc seems to be saying that, in a split brain condition, two masters > will exist, hence the attempt to determine LAN connectivity. But, if the > comms issue was fleeting, and now comms are "good", what then? We have two > masters, good comms, who wins? Does Cluster Services retain knowledge of > who was the previous Master, and it wins? > > The root of this is an infrequent (rare, actually) reboot of non master > nodes. OES1, Linux, 2 node cluster. > > joe a. > > >>> "John Cunningham" 04/30/09 9:47 AM >>> > Split brain condition? > > I keep referring back to TID 10053882 when it comes to this stuff. > > http://support.novell.com/docs/Tids/Solutions/10053882.html > > >>> On 4/30/2009 at 10:42 AM, Bill Brush wrote: > I think you have it right if my sessions at Brainshare are coming back > to me correctly. > > Bill > > On Thu, Apr 30, 2009 at 8:01 AM, joea at j4computers.com > wrote: > > Somebody know cluster services? > > > > In a 2 node cluster, docs seem to say a special condition exists, (when > there is a loss of heartbeat) in that, since there cannot be a "majority", > nor can the Master node be assumed "good", then, if network connectivity can > be determined, the "bad" node eats a poison pill. > > > > However, docs do not state what happens when connectivity *cannot* be > determined. Does the Master then always win? From experience, it would seem > so, as we have had instances of the non master node rebooting, apparently > spontaneously. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- My blog: http://www.antioxidantalley.com/blog LinkedIn Profile: http://www.linkedin.com/in/jbchristensen Facebook Profile: http://profile.to/jbchristensen/ Follow me on Twitter: http://twitter.com/jbchristensen _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From douglasbecker at cox.net Sun May 3 21:53:28 2009 From: douglasbecker at cox.net (Douglas Becker) Date: Sun, 3 May 2009 16:53:28 -0400 (EDT) Subject: Whither Novell Message-ID: <24479978.14320.1241384008618.JavaMail.douglasbecker@127.0.0.1> Marty, I know where you are coming from. I work for a contractor and the accounting system that we use and the Fixed Asset program that we use are strictly Windows only (1 of them REQUIRES SQL). I am forced more and more to put up Windows servers and I am loosing the battle to keep GroupWise (all the built-in mail functions require that you have Outlook installed on your computer, and the upper level do not want to upgrade to GW8 so that I can do that) so I may be forced to drop Novell altogether and go with Windows. Believe me I don't want to, but it seems more and more that I am being forced to. Douglas Becker douglasbecker at cox.net On Fri, May 1, 2009 at 4:48 PM , Matt Smith wrote: >>>> On 4/30/2009 at 11:02 PM, in message > , Randy Grein > wrote: >> Sorry Marty, but I can't buy the logic you're using. Can you explain > >> more? >> >> Linux has steadily increased in market share, eating mostly into >> Netware (but, AFAICT Windows as well) so that it's somewhere between > >> 25-35% of new installs. Support is available from multiple vendors >> (open source, SUSE, Red Hat, Debian, Ubuntu, etc) so what's the real > >> problem? >> >> Keep in mind the distinction between security and profit. Security >> calls for following the herd. Profit calls for striking out on your > >> own (correct) path. How much do you trust your own analysis? > > I don't want to "gang up" on you, Marty, but my first response was > pretty much the same as Randy's here. We do have several windows > servers running as application servers. However, I've found lately > that > we were installing Windows to run these application not because they > were Windows only, but because the sales people pitched it as a > Windows > application to the power-that-be. Turns out that if we'd asked, we > could have been running some of this stuff on Linux after all. I'm > particularly burned by the recent purchase of our HR system. The > company that owns the product actually develops on Linux and ports it > over to Windows! > > I'm not being ingenuous. Windows dominates the market. Still, my > boss > loves to brag when he hears the latest school system that lost a day > or > two's work to the latest internet nasty while we soldier on without a > hitch. > > Besides, SUSE is linux. I figure if I have to, we'll just switch to a > different flavor. I don't expect my experience to be quite as rich as > what I get with Novell, but I also don't feel like I'm wasting my time > now by any means. > > Each to their own though. > > -Matt > -- > > Matt Smith Network Technology Specialist > Oconee County School System, Oconee County, Georgia > Office of Instruction and Technology 706-769-5685 x1314 > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From cmangiarelli at gmail.com Mon May 4 18:17:12 2009 From: cmangiarelli at gmail.com (Christopher Mangiarelli) Date: Mon, 4 May 2009 13:17:12 -0400 Subject: NW6.0 to NW6.5 in-place upgrade - "rollback" option? In-Reply-To: <68b791330905010404ja526ea7x5c5e65822e9f26fc@mail.gmail.com> References: <68b791330904300507s1198ff5as8353fa4d8016ae84@mail.gmail.com> <68b791330904300629p6e1e6cddkcb9daf767ac89bb9@mail.gmail.com> <49F97FF9020000750003152C@inet.eastcobbgroup.com> <68b791330904300858vcc49563s331f8ce8b9f3fe8d@mail.gmail.com> <49F9E9D7.6050902@calfrye.com> <68b791330904301227k71f72e2ch8de8c3bd92c2c9e4@mail.gmail.com> <49F9D571020000750004C47F@inet.eastcobbgroup.com> <68b791330904301934q788d6516p8a7fde1137a4bb7c@mail.gmail.com> <393865BA-0F21-4F5D-B5AB-DAF43D170774@gmail.com> <68b791330905010404ja526ea7x5c5e65822e9f26fc@mail.gmail.com> Message-ID: Most recently I did it from the server in the X display, but I've also done it across the wire using NWDEPLOY. On Fri, May 1, 2009 at 7:04 AM, Peter Van Lone wrote: > On Fri, May 1, 2009 at 12:28 AM, Christopher Mangiarelli > wrote: > > I've done several in place upgrades from nw6 to nw65. Used a sp6 overlay > > cd. > > which method did you use? "Down server" upgrade where you boot to the > ne 65 overlay CD? > > > Like joe said, make sure you have space and Edir is healthy. Obviously > > backups are important. Get the environment to edir 8.7.3.10b and nw65sp7 > > after all is well. Can also go to sp8 too. Both are working well for me. > > thnx! > -- Christopher Mangiarelli cmangiarelli at gmail.com From bjonkman at sobac.com Mon May 4 19:38:12 2009 From: bjonkman at sobac.com (Bob Jonkman) Date: Mon, 04 May 2009 14:38:12 -0400 Subject: Whither Novell In-Reply-To: <49FA3C15.2000109@toombs.com> References: <49FA3C15.2000109@toombs.com> Message-ID: <49FF3614.2080204@sobac.com> A non-profit organization is usually an insufficient-funds organization too. I would think it all the more important to migrate to a platform that has long-term support, and to avoid locking in with one particular vendor (neither Microsoft nor Novell). It's an added benefit that most of the Linux OS and applications are available at no cost to purchase. Support costs are less too, and you have the choice of support vendor, unlike for commercial OSes. You give a perfect example of suffering from vendor lock-in: "Netware has left me and the fact that Novell is no longer producing new versions was their choice, not mine." Microsoft is no different. MS-Server 2008 will someday reach end-of-life, and you'll be forced to buy all new hardware in order to run the newest offerings. I'll bet that the bulk of your computing needs are for standard office productivity: Payroll, HR, CRM, and the standard desktop applications of word processor, spreadsheet, e-mail and maybe Web browsing. I'll bet that you can meet all your computing needs with Linux-based software. Yes, there is some cost to switching platforms, but I'll also bet that the cost to switch to Linux today is less than the cost of upgrading to Microsoft licenses, not to mention the Microsoft licence renewal costs year after year. --Bob. Marty Toombs wrote: > I have a slightly different take on the Netware issue and what > versions to compare to what. > > I'd like to compare Netware 7 to Windows Server 2008. But I can't do > that. > > I feel that Netware has left me and the fact that Novell is no longer > producing new versions was their choice, not mine. I'm not leaving > Novell, they left me. > > I'm the IS director for a small non-profit. In today's environment I > can (and have) argued for Netware for all the reasons everyone on this > list knows. At this point, those reasons don't exist for Linux. At the > same time I'm dealing with more and more applications which are > Windows only. Our payroll program used to run on Netware, now it's > Windows only, etc. I've been forced to set up Windows servers next to > my Netware boxes just to keep doing what I'm doing. > > As I look to the future, I think it would be irresponsible for me to > go to SUSE servers when I can't be certain that proper support for > them will be available. I have to think of the agency's future and not > just my wishes. > > So the end of Netware for me has left me with just one choice -- how > fast do we move to Windows. > > My current plan calls for the last Netware server to come down some > time next year. I will not enjoy that when it happens. > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell -- Bob Jonkman http://sobac.com/sobac/ SOBAC Microcomputer Services Voice: +1-519-669-0388 6 James Street, Elmira ON Canada N3B 1L5 Cel: +1-519-635-9413 Software --- Office & Business Automation --- Consulting From RGrein at tpchd.org Mon May 4 20:26:51 2009 From: RGrein at tpchd.org (Randy Grein) Date: Mon, 04 May 2009 12:26:51 -0700 Subject: Whither Novell In-Reply-To: <49FF3614.2080204@sobac.com> References: <49FA3C15.2000109@toombs.com> <49FF3614.2080204@sobac.com> Message-ID: <49FEDF0B0200007200032DB5@health-mail2.tpchd.org> Absolutely right Bob. A good-sized company here in Tacoma was featured in Computerworld last year; their 3 year Microsoft licensing costs paid for a full migration to Apple - workstations and servers, with, IIRC $1.2m to spare. The glue used by apple is OpenLDAP which is (of course) available on many other platforms. Point is, the cost savings are there for many, if not most organizations. Randy Grein Sr. Network Engineer >>> Bob Jonkman 5/4/2009 11:38 AM >>> I'll bet that the bulk of your computing needs are for standard office productivity: Payroll, HR, CRM, and the standard desktop applications of word processor, spreadsheet, e-mail and maybe Web browsing. I'll bet that you can meet all your computing needs with Linux-based software. Yes, there is some cost to switching platforms, but I'll also bet that the cost to switch to Linux today is less than the cost of upgrading to Microsoft licenses, not to mention the Microsoft licence renewal costs year after year. --Bob. ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From douglasbecker at cox.net Mon May 4 21:07:56 2009 From: douglasbecker at cox.net (Douglas Becker) Date: Mon, 4 May 2009 16:07:56 -0400 (EDT) Subject: Whither Novell Message-ID: <13934597.16097.1241467676774.JavaMail.douglasbecker@127.0.0.1> But I think that you are missing the point with what he was saying. He said that some of his apps were windows only. I have the same problem. I cannot get that program in a Linux version. Therefore I am forced to switch to windows server. And it may be true that the cost to upgrade to Linux is less, but if it will not run the programs that I have to run, why would I use it at all? Douglas Becker douglasbecker at cox.net On Mon, May 4, 2009 at 4:03 PM , Randy Grein wrote: > Absolutely right Bob. A good-sized company here in Tacoma was featured > in Computerworld last year; their 3 year Microsoft licensing costs > paid for a full migration to Apple - workstations and servers, with, > IIRC $1.2m to spare. The glue used by apple is OpenLDAP which is (of > course) available on many other platforms. Point is, the cost savings > are there for many, if not most organizations. > > > > > Randy Grein > Sr. Network Engineer >>>> Bob Jonkman 5/4/2009 11:38 AM >>> > I'll bet that the bulk of your computing needs are for standard office > productivity: Payroll, HR, CRM, and the standard desktop applications > of word processor, spreadsheet, e-mail and maybe Web browsing. I'll > bet that you can meet all your computing needs with Linux-based > software. Yes, there is some cost to switching platforms, but I'll > also bet that the cost to switch to Linux today is less than the cost > of upgrading to Microsoft licenses, not to mention the Microsoft > licence renewal costs year after year. > > --Bob. > > > ************************************************************************************* > This e-mail and any attachments may contain confidential and > privileged information. It has been scanned for viruses. If you are > not the intended recipient, please notify the sender immediately by > return e-mail, delete this e-mail and destroy any copies. Any > dissemination, use, review, disclosure, or distribution of this > information by a person other than the intended recipient is > unauthorized and may be illegal. > > ************************************************************************************** > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Mon May 4 21:18:43 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Mon, 04 May 2009 16:18:43 -0400 Subject: Linux Swap "preallocate" or not? Message-ID: <49FF1554020000850005FC54@FS-LIN-OES> This was touched on a while back, but needs a revisit. At least by me. Does, or does not, Linux (OES1, in this case) "pre-allocate" swap space that has been defined? And what, pray tell, might that actually mean? Have a couple of boxes that appear to be using up swap. Top (q, f), swapon -s, vmstat, all seem to indicate that "swap" space is growing. That is, more of the defined "swap" is being "used". Have created additional swap files, with same priority as swap partition, and those are slowly being consumed as well. I can see the file size is increasing. Yet, I am being told this does not mean the swap is "in use", but has merely been "pre-allocated". The "" are being used here mainly to indicate uncertainty about terminology. Mainly. Also, if "swapping" were actually happening, to the extent on might suspect, that disk thrashing would be reducing performance and increasing user complaint. Yet, if "swap" is not being used, why is the "swap file" size increasing? joe a. From cal.frye at oberlin.edu Mon May 4 21:43:43 2009 From: cal.frye at oberlin.edu (Cal Frye) Date: Mon, 04 May 2009 16:43:43 -0400 Subject: Whither Novell In-Reply-To: <13934597.16097.1241467676774.JavaMail.douglasbecker@127.0.0.1> References: <13934597.16097.1241467676774.JavaMail.douglasbecker@127.0.0.1> Message-ID: <49FF537F.8000809@oberlin.edu> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Douglas Becker wrote: > But I think that you are missing the point with what he was saying. He > said that some of his apps were windows only. I have the same problem. > I cannot get that program in a Linux version. Therefore I am forced to > switch to windows server. And it may be true that the cost to upgrade > to Linux is less, but if it will not run the programs that I have to > run, why would I use it at all? I'm having some difficulties with this argument. I've used two whole third-party applications that ran natively on Netware. Everything else was file/print or NetMail (and backup software, but I'm not counting non-user applications here). I've almost never run a Netware shop that didn't also have a Windows server, but the file/print performance issues have always made Netware a good choice. You can choose to standardize on a single server platform and minimize administrator time and hassle, or you can attempt to serve the users with best-of-breed choices for the tasks at hand. Both are valid and real choices, but let's acknowledge that the monoculture may not always be the best choice from the users' perspective. - -- Celebrating the 150th anniversary of the publication of the Origin of Species. - -- Cal Frye, Network Administrator, Oberlin College Mudd Library, x.56930 -- CIT will NEVER ask you for your password! www.calfrye.com, www.pitalabs.com "It's wonderful how much you can accomplish when you decide you don't need to take the credit for it." -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (Darwin) iEYEARECAAYFAkn/U38ACgkQcZlA4wu9pSCkZgCeJZa0glbEP9TrSkul1sOuy6YU arAAnidKOueS8T6e9XUmWZZSOfjILaTk =55hp -----END PGP SIGNATURE----- From RGrein at tpchd.org Mon May 4 22:09:17 2009 From: RGrein at tpchd.org (Randy Grein) Date: Mon, 04 May 2009 14:09:17 -0700 Subject: Whither Novell In-Reply-To: <13934597.16097.1241467676774.JavaMail.douglasbecker@127.0.0.1> References: <13934597.16097.1241467676774.JavaMail.douglasbecker@127.0.0.1> Message-ID: <49FEF70D0200007200032DDB@health-mail2.tpchd.org> Not knowing your requirement it's hard to say. It is common for a particular application vendor to deploy only on one platform but there are almost always substitutes. The quality of the substitute may be unacceptable for the situation. There may be legal requirements, or political (the owner will accept ONLY Autocad or ONLY Microsoft Office, not trubocad or Open Office). The amount of work required to satisfy such a requirement may be too much (running all applications on linux workstations in a virtual environment is a good example - possible but inefficient). My point wasn't to say that you missed something obvious, just that we tend to get locked up in a one-solution mindset, especially when vendor lockin happens. It's why economists do not talk about needs, only wants. Even air will be bartered if something else more valuable is needed. Randy Grein Sr. Network Engineer >>> Douglas Becker 5/4/2009 1:07 PM >>> But I think that you are missing the point with what he was saying. He said that some of his apps were windows only. I have the same problem. I cannot get that program in a Linux version. Therefore I am forced to switch to windows server. And it may be true that the cost to upgrade to Linux is less, but if it will not run the programs that I have to run, why would I use it at all? Douglas Becker douglasbecker at cox.net On Mon, May 4, 2009 at 4:03 PM , Randy Grein wrote: > Absolutely right Bob. A good-sized company here in Tacoma was featured > in Computerworld last year; their 3 year Microsoft licensing costs > paid for a full migration to Apple - workstations and servers, with, > IIRC $1.2m to spare. The glue used by apple is OpenLDAP which is (of > course) available on many other platforms. Point is, the cost savings > are there for many, if not most organizations. > > > > > Randy Grein > Sr. Network Engineer >>>> Bob Jonkman 5/4/2009 11:38 AM >>> > I'll bet that the bulk of your computing needs are for standard office > productivity: Payroll, HR, CRM, and the standard desktop applications > of word processor, spreadsheet, e-mail and maybe Web browsing. I'll > bet that you can meet all your computing needs with Linux-based > software. Yes, there is some cost to switching platforms, but I'll > also bet that the cost to switch to Linux today is less than the cost > of upgrading to Microsoft licenses, not to mention the Microsoft > licence renewal costs year after year. > > --Bob. > > > ************************************************************************************* > This e-mail and any attachments may contain confidential and > privileged information. It has been scanned for viruses. If you are > not the intended recipient, please notify the sender immediately by > return e-mail, delete this e-mail and destroy any copies. Any > dissemination, use, review, disclosure, or distribution of this > information by a person other than the intended recipient is > unauthorized and may be illegal. > > ************************************************************************************** > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From jrd at netlab1.oucs.ox.ac.uk Mon May 4 22:26:10 2009 From: jrd at netlab1.oucs.ox.ac.uk (jrd) Date: Mon, 04 May 2009 22:26:10 +0100 Subject: Linux Swap "preallocate" or not? In-Reply-To: <49FF1554020000850005FC54@FS-LIN-OES> References: <49FF1554020000850005FC54@FS-LIN-OES> Message-ID: <49FF5D72.2090203@netlab1.oucs.ox.ac.uk> joea at j4computers.com wrote: > This was touched on a while back, but needs a revisit. At least by me. > > Does, or does not, Linux (OES1, in this case) "pre-allocate" swap space that has been defined? And what, pray tell, might that actually mean? > > Have a couple of boxes that appear to be using up swap. Top (q, f), swapon -s, vmstat, all seem to indicate that "swap" space is growing. That is, more of the defined "swap" is being "used". Have created additional swap files, with same priority as swap partition, and those are slowly being consumed as well. I can see the file size is increasing. > > Yet, I am being told this does not mean the swap is "in use", but has merely been "pre-allocated". The "" are being used here mainly to indicate uncertainty about terminology. Mainly. > > Also, if "swapping" were actually happening, to the extent on might suspect, that disk thrashing would be reducing performance and increasing user complaint. > > Yet, if "swap" is not being used, why is the "swap file" size increasing? > > joe a. > ------------ Joe, That does seem odd to me. One can create new swap files on the fly and use swapon as you have. Normally we create a swap partition of reasonable size and see if it is consumed. By then, as you also note, the system ought to become sluggish. I create 256MB or 512MB swap partitions, depending upon expected system usage, and they rarely consume that much space. So, first, what does "top" say about swap usage? Then is there something on that system which keeps asking for more and more memory? We can use the "ps aux" command to see both real and virtual memory usage. Some turning off of services one by one might reveal the culprit. Below are examples from one of my (64 bit OES2 SP1) machines: > $ top > top - 22:19:39 up 13 days, 10:00, 1 user, load average: 0.14, 0.15, 0.16 > Tasks: 296 total, 1 running, 295 sleeping, 0 stopped, 0 zombie > Cpu(s): 1.6%us, 1.6%sy, 0.0%ni, 95.3%id, 1.6%wa, 0.0%hi, > 0.0%si, 0.0%st > Mem: 5077084k total, 5033808k used, 43276k free, 4k buffers > Swap: 530104k total, 164k used, 529940k free, 2390096k cached > $ ps aux (VSZ is virtual mem, RSS is real/physical > mem) > USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND > root 1 0.0 0.0 800 308 ? S Apr21 0:01 init [3] > ... > root 1991 0.0 0.0 0 0 ? S< Apr21 0:00 > [xfssyncd] > root 2064 0.0 0.0 0 0 ? S< Apr21 0:00 [kauditd] > root 2371 0.0 0.2 53400 15020 ? Ssl Apr21 0:01 > /usr/bin/micasad-init /usr/bin/micasad.e > root 2381 0.0 0.0 4304 512 ? Ss Apr21 0:00 > /usr/sbin/irqbalance > root 2387 0.0 0.0 23380 1176 ? Ss Apr21 0:00 > /sbin/resmgrd > root 2415 0.0 0.0 2680 524 ? Ss Apr21 0:00 > /sbin/acpid > 100 2534 0.0 0.0 25280 1192 ? Ss Apr21 0:00 > /usr/bin/dbus-daemon --system > root 2546 0.0 0.0 13384 3424 ? Ss Apr21 0:00 > /usr/sbin/hald --daemon=yes --retain-pri > root 2719 0.0 0.0 6072 684 ? S Apr21 0:00 > hald-addon-acpi > root 2979 0.1 0.0 6068 660 ? S Apr21 19:30 > hald-addon-storage > root 3098 0.0 0.0 0 0 ? S Apr23 0:00 [MPK > Thread] > ... > root 3126 0.0 0.0 0 0 ? S Apr23 0:00 [MPK > Thread] > root 3133 0.0 0.0 2736 456 ? Ss Apr21 0:00 > startpar -f -- mailman > root 3134 0.0 0.0 8360 900 ? Ss Apr21 0:21 > /sbin/syslog-ng > $ vmstat > procs -----------memory---------- ---swap-- -----io---- -system-- > -----cpu------ > r b swpd free buff cache si so bi bo in cs us > sy id wa st > 0 0 164 43380 4 2392152 0 0 15 25 7 16 3 > 1 95 0 0 > # swapon -s > Filename Type Size > Used Priority > /dev/sdb1 partition 530104 164 -1 Joe D. From joea at j4computers.com Mon May 4 23:29:32 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Mon, 04 May 2009 18:29:32 -0400 Subject: Linux Swap "preallocate" or not? In-Reply-To: <49FF5D72.2090203@netlab1.oucs.ox.ac.uk> References: <49FF1554020000850005FC54@FS-LIN-OES> <49FF5D72.2090203@netlab1.oucs.ox.ac.uk> Message-ID: <49FF340C020000850005FC5E@FS-LIN-OES> >>> On 5/4/2009 at 5:26 PM, jrd wrote: > joea at j4computers.com wrote: >> This was touched on a while back, but needs a revisit. At least by me. >> >> Does, or does not, Linux (OES1, in this case) "pre-allocate" swap space that > has been defined? And what, pray tell, might that actually mean? >> >> Have a couple of boxes that appear to be using up swap. Top (q, f), swapon > -s, vmstat, all seem to indicate that "swap" space is growing. That is, > more of the defined "swap" is being "used". Have created additional swap > files, with same priority as swap partition, and those are slowly being > consumed as well. I can see the file size is increasing. >> >> Yet, I am being told this does not mean the swap is "in use", but has merely > been "pre-allocated". The "" are being used here mainly to indicate > uncertainty about terminology. Mainly. >> >> Also, if "swapping" were actually happening, to the extent on might suspect, > that disk thrashing would be reducing performance and increasing user > complaint. >> >> Yet, if "swap" is not being used, why is the "swap file" size increasing? >> >> joe a. >> > ------------ > Joe, > That does seem odd to me. One can create new swap files on the > fly and use swapon as you have. > Normally we create a swap partition of reasonable size and see if it is > consumed. By then, as you also note, > the system ought to become sluggish. I create 256MB or 512MB swap > partitions, depending upon expected > system usage, and they rarely consume that much space. > So, first, what does "top" say about swap usage? Then is there > something on that system which > keeps asking for more and more memory? We can use the "ps aux" command > to see both real and > virtual memory usage. Some turning off of services one by one might > reveal the culprit. . . . I cannot provide the top output, right now, as I have left the building. But, from memory, ahem, sorry, recollection . . .top tells me that the GroupWise Post Office Agents, are consuming the majority of the swap. Or so it seems, having typed, f, at the top screen, and p (?) at the "menu" screen to turn on swap display. The amount of swap that is indicated in use, varies a bit, but is very near 1 GB for each poa process. The system has 4 GB of RAM, and now has 4GB of swap. joe a. From MollardM at mbc.qld.edu.au Tue May 5 02:57:14 2009 From: MollardM at mbc.qld.edu.au (Michael Mollard) Date: Tue, 05 May 2009 11:57:14 +1000 Subject: Acrobat and long iPrint printer names Message-ID: <4A00299A.E927.0018.0@mbc.qld.edu.au> Hi all, I just want to see if anyone else has come across this problem. We use iPrint for our printing. Our printer names are (eg:) ipp://iprint.ourserver.edu.au/printername-colour. In Adobe Acrobat Reader 9, when I open the Printer dialog to print, the right hand side of the printer name is truncated, which in our case means I can't see whether the printer is Black or Colour. Other applications seem to use "Printername on ipp://servername" which means we can see the printer description. Has anyone found this, or solved it? Thanks. Michael Mollard Network Administrator Moreton Bay College mollardm at mbc.qld.edu.au http://www.mbc.qld.edu.au Ph: (Direct) 07 3907 5712 / (Mob) 0417 631 801 Fax: 07 3390 8919 ( http://www.mbc.qld.edu.au ) -------------------------------------------------------------------------------- Disclaimer: Whilst every attempt has been made to ensure that material contained in this email is free from computer viruses or other defects, the attached files are provided, and may only be used, on the basis that the user assumes all responsibility for use of the material transmitted. This email is intended only for the use of the individual or entity names above and may contain information that is confidential and privileged. If you are not the intended recipient, please note that any dissemination, distribution or copying of this email is strictly prohibited. If you have received this email in error, please notify us immediately by return email or telephone 07 3390 8555 and destroy the original message. The contents of this message are provided without responsibility in law for their accuracy or otherwise, and without assumption of a duty of care by the School. -------------------------------------------------------------------------------- From randygrein at comcast.net Tue May 5 04:21:29 2009 From: randygrein at comcast.net (Randy Grein) Date: Mon, 4 May 2009 20:21:29 -0700 Subject: Acrobat and long iPrint printer names In-Reply-To: <4A00299A.E927.0018.0@mbc.qld.edu.au> References: <4A00299A.E927.0018.0@mbc.qld.edu.au> Message-ID: <470F6BA1-F3CA-4F18-A2B2-E7816BA32091@comcast.net> Found this on several applications - gotta love the standard nonstandard selection boxes! Sadly I know of no solutions except short names. Randy Grein, Master CNE, CCNA On May 4, 2009, at 6:57 PM, Michael Mollard wrote: > Hi all, > I just want to see if anyone else has come across this problem. > We use iPrint for our printing. Our printer names are (eg:) ipp:// > iprint.ourserver.edu.au/printername-colour. In Adobe Acrobat Reader > 9, when I open the Printer dialog to print, the right hand side of > the printer name is truncated, which in our case means I can't see > whether the printer is Black or Colour. > > Other applications seem to use "Printername on ipp://servername" > which means we can see the printer description. > > Has anyone found this, or solved it? > > Thanks. > > > Michael Mollard > Network Administrator > Moreton Bay College > mollardm at mbc.qld.edu.au > http://www.mbc.qld.edu.au > Ph: (Direct) 07 3907 5712 / (Mob) 0417 631 801 > Fax: 07 3390 8919 ( http://www.mbc.qld.edu.au ) > > > -------------------------------------------------------------------------------- > Disclaimer: Whilst every attempt has been made to ensure that > material contained in this email is free from computer viruses or > other defects, the attached files are provided, and may only be > used, on the basis that the user assumes all responsibility for use > of the material transmitted. This email is intended only for the use > of the individual or entity names above and may contain information > that is confidential and privileged. If you are not the intended > recipient, please note that any dissemination, distribution or > copying of this email is strictly prohibited. If you have received > this email in error, please notify us immediately by return email or > telephone 07 3390 8555 and destroy the original message. The > contents of this message are provided without responsibility in law > for their accuracy or otherwise, and without assumption of a duty of > care by the School. > -------------------------------------------------------------------------------- > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From douglasbecker at cox.net Tue May 5 17:26:19 2009 From: douglasbecker at cox.net (Douglas Becker) Date: Tue, 5 May 2009 12:26:19 -0400 (EDT) Subject: Whither Novell Message-ID: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> I agree with you there. My situation is that our accounting vendor stopped supporting and developing for Netware. My reaction was to find another accounting package that did. Unfortunately, the CFO "WILL NOT" (her words) do a migration of the accounting data to another package. Therefore I was stuck with having to get another server (hardware and all) just to put the accounting package on it. The same thing happened again with the fixed asset package. Now I have a server that does nothing but file and print sharing using Netware, one Netware server doing GW and two Windows server for the other stuff. I am using GW 6.5 and to get the pipe to allow the use of Outlook, I need to upgrade to 8, and they won't do that because there is a chance we might just move everything to Windows. I have some say in that, but unfortunately it does not carry a lot of weight. Aside from allowing me to keep GW, what would moving to Linux do for me? Douglas Becker douglasbecker at cox.net On Mon, May 4, 2009 at 12:18 PM , Randy Grein wrote: > Not knowing your requirement it's hard to say. It is common for a > particular application vendor to deploy only on one platform but there > are almost always substitutes. The quality of the substitute may be > unacceptable for the situation. There may be legal requirements, or > political (the owner will accept ONLY Autocad or ONLY Microsoft > Office, not trubocad or Open Office). The amount of work required to > satisfy such a requirement may be too much (running all applications > on linux workstations in a virtual environment is a good example - > possible but inefficient). My point wasn't to say that you missed > something obvious, just that we tend to get locked up in a > one-solution mindset, especially when vendor lockin happens. It's why > economists do not talk about needs, only wants. Even air will be > bartered if something else more valuable is needed. > > > > > > Randy Grein > Sr. Network Engineer >>>> Douglas Becker 5/4/2009 1:07 PM >>> > But I think that you are missing the point with what he was saying. > He said that some of his apps were windows only. I have the same > problem. I cannot get that program in a Linux version. Therefore I am > forced to switch to windows server. And it may be true that the cost > to upgrade to Linux is less, but if it will not run the programs that > I have to run, why would I use it at all? > > Douglas Becker > > douglasbecker at cox.net > > On Mon, May 4, 2009 at 4:03 PM , Randy Grein wrote: > >> Absolutely right Bob. A good-sized company here in Tacoma was >> featured in Computerworld last year; their 3 year Microsoft licensing >> costs paid for a full migration to Apple - workstations and servers, >> with, IIRC $1.2m to spare. The glue used by apple is OpenLDAP which >> is (of course) available on many other platforms. Point is, the cost >> savings are there for many, if not most organizations. >> >> >> >> >> Randy Grein >> Sr. Network Engineer >>>>> Bob Jonkman 5/4/2009 11:38 AM >>> >> I'll bet that the bulk of your computing needs are for standard >> office productivity: Payroll, HR, CRM, and the standard desktop >> applications of word processor, spreadsheet, e-mail and maybe Web >> browsing. I'll bet that you can meet all your computing needs with >> Linux-based software. Yes, there is some cost to switching >> platforms, but I'll also bet that the cost to switch to Linux today >> is less than the cost of upgrading to Microsoft licenses, not to >> mention the Microsoft licence renewal costs year after year. >> >> --Bob. >> >> >> >> ************************************************************************************* >> This e-mail and any attachments may contain confidential and >> privileged information. It has been scanned for viruses. If you are >> not the intended recipient, please notify the sender immediately by >> return e-mail, delete this e-mail and destroy any copies. Any >> dissemination, use, review, disclosure, or distribution of this >> information by a person other than the intended recipient is >> unauthorized and may be illegal. >> >> >> ************************************************************************************** >> >> >> _______________________________________________ >> Novell mailing list >> Novell at netlab1.oucs.ox.ac.uk >> http://netlab1.usu.edu/mailman/listinfo/novell > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > > ************************************************************************************* > This e-mail and any attachments may contain confidential and > privileged information. It has been scanned for viruses. If you are > not the intended recipient, please notify the sender immediately by > return e-mail, delete this e-mail and destroy any copies. Any > dissemination, use, review, disclosure, or distribution of this > information by a person other than the intended recipient is > unauthorized and may be illegal. > > ************************************************************************************** > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From Simon.Shilton at acustica.co.uk Tue May 5 17:42:27 2009 From: Simon.Shilton at acustica.co.uk (Simon Shilton) Date: Tue, 05 May 2009 17:42:27 +0100 Subject: Whither Novell In-Reply-To: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> References: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> Message-ID: <4A007A83020000AB0001ACC7@dylan.trident.acustica.co.uk> from a purely practical perspective, I would suggest you pick which fight you want to fight Personally I would be tempted to move the NetWare file/print services to Windows, and upgrade to GW8 and run the agents on Windows they get the Windows they want, you keep the GW you want you support one server OS, and you don't have to shell out a fortune for Exchange or the pain of migration to Exchange, and you get the nice new GW8 full text indexing of PDF files which doesn't work on NW anyway I love using NetWare, but realistically if you haven't got it virtualised, you have a real struggle getting in onto brand new hardware, and it ain't going nowhere. OES2sp1 on SLES10sp2 is almost a solid (and possible more so in some aspects) than NW6.5sp8 but requires a whole bunch of new skills, and some mind share from the higher ups which are probably a challenge. Give them what they want, which is Windows server OS, and keep GW. Maybe after Windows servers give you grief Linux can be introduced as the lower cost and more robust solution. :-) best of luck Simon >>> On 05 May 2009 at 17:26, in message <30899801.19161.1241540779042.JavaMail.douglasbecker at 127.0.0.1>, Douglas Becker wrote: I agree with you there. My situation is that our accounting vendor stopped supporting and developing for Netware. My reaction was to find another accounting package that did. Unfortunately, the CFO "WILL NOT" (her words) do a migration of the accounting data to another package. Therefore I was stuck with having to get another server (hardware and all) just to put the accounting package on it. The same thing happened again with the fixed asset package. Now I have a server that does nothing but file and print sharing using Netware, one Netware server doing GW and two Windows server for the other stuff. I am using GW 6.5 and to get the pipe to allow the use of Outlook, I need to upgrade to 8, and they won't do that because there is a chance we might just move everything to Windows. I have some say in that, but unfortunately it does not carry a lot of weight. Aside from allowing me to keep GW, what would moving to Linux do for me? Douglas Becker douglasbecker at cox.net On Mon, May 4, 2009 at 12:18 PM , Randy Grein wrote: > Not knowing your requirement it's hard to say. It is common for a > particular application vendor to deploy only on one platform but there > are almost always substitutes. The quality of the substitute may be > unacceptable for the situation. There may be legal requirements, or > political (the owner will accept ONLY Autocad or ONLY Microsoft > Office, not trubocad or Open Office). The amount of work required to > satisfy such a requirement may be too much (running all applications > on linux workstations in a virtual environment is a good example - > possible but inefficient). My point wasn't to say that you missed > something obvious, just that we tend to get locked up in a > one-solution mindset, especially when vendor lockin happens. It's why > economists do not talk about needs, only wants. Even air will be > bartered if something else more valuable is needed. > > > > > > Randy Grein > Sr. Network Engineer >>>> Douglas Becker 5/4/2009 1:07 PM >>> > But I think that you are missing the point with what he was saying. > He said that some of his apps were windows only. I have the same > problem. I cannot get that program in a Linux version. Therefore I am > forced to switch to windows server. And it may be true that the cost > to upgrade to Linux is less, but if it will not run the programs that > I have to run, why would I use it at all? > > Douglas Becker > > douglasbecker at cox.net > > On Mon, May 4, 2009 at 4:03 PM , Randy Grein wrote: > >> Absolutely right Bob. A good-sized company here in Tacoma was >> featured in Computerworld last year; their 3 year Microsoft licensing >> costs paid for a full migration to Apple - workstations and servers, >> with, IIRC $1.2m to spare. The glue used by apple is OpenLDAP which >> is (of course) available on many other platforms. Point is, the cost >> savings are there for many, if not most organizations. >> >> >> >> >> Randy Grein >> Sr. Network Engineer >>>>> Bob Jonkman 5/4/2009 11:38 AM >>> >> I'll bet that the bulk of your computing needs are for standard >> office productivity: Payroll, HR, CRM, and the standard desktop >> applications of word processor, spreadsheet, e-mail and maybe Web >> browsing. I'll bet that you can meet all your computing needs with >> Linux-based software. Yes, there is some cost to switching >> platforms, but I'll also bet that the cost to switch to Linux today >> is less than the cost of upgrading to Microsoft licenses, not to >> mention the Microsoft licence renewal costs year after year. >> >> --Bob. >> >> >> >> ************************************************************************************* >> This e-mail and any attachments may contain confidential and >> privileged information. It has been scanned for viruses. If you are >> not the intended recipient, please notify the sender immediately by >> return e-mail, delete this e-mail and destroy any copies. Any >> dissemination, use, review, disclosure, or distribution of this >> information by a person other than the intended recipient is >> unauthorized and may be illegal. >> >> >> ************************************************************************************** >> >> >> _______________________________________________ >> Novell mailing list >> Novell at netlab1.oucs.ox.ac.uk >> http://netlab1.usu.edu/mailman/listinfo/novell > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > > ************************************************************************************* > This e-mail and any attachments may contain confidential and > privileged information. It has been scanned for viruses. If you are > not the intended recipient, please notify the sender immediately by > return e-mail, delete this e-mail and destroy any copies. Any > dissemination, use, review, disclosure, or distribution of this > information by a person other than the intended recipient is > unauthorized and may be illegal. > > ************************************************************************************** > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell -- This email was Anti Virus checked by Astaro Security Gateway. http://www.astaro.com From petervl at gmail.com Tue May 5 18:14:40 2009 From: petervl at gmail.com (Peter Van Lone) Date: Tue, 5 May 2009 12:14:40 -0500 Subject: Whither Novell In-Reply-To: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> References: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> Message-ID: <68b791330905051014q47548b9ao76f4cb6865d58f2a@mail.gmail.com> On Tue, May 5, 2009 at 11:26 AM, Douglas Becker wrote: Therefore I was > stuck with having to get another server (hardware and all) just to put the > accounting package on it. ?The same thing happened again with the fixed > asset package. ?Now I have a server that does nothing but file and print > sharing using Netware, one Netware server doing GW and two Windows server > for the other stuff. this is exactly why you should be virtualized. Then you can add/remove servers as needed, while investing in only one (or two for redundancy) physical server devices. You can get started with no shared storage -- and if that is the case then ESXi (which is free) is to my mind the perfect way to go. >?I am using GW 6.5 and to get the pipe to allow the use > of Outlook, I need to upgrade to 8, and they won't do that because there is > a chance we might just move everything to Windows. ?I have some say in that, > but unfortunately it does not carry a lot of weight. ?Aside from allowing me > to keep GW, what would moving to Linux do for me? keep you on a supported NCP solution that is familiar to your users and requires no new training or changes for them? Prevents you from being locked in by any one vendor? Allows you to leverage iPrint? P From RGrein at tpchd.org Tue May 5 18:18:48 2009 From: RGrein at tpchd.org (Randy Grein) Date: Tue, 05 May 2009 10:18:48 -0700 Subject: Whither Novell In-Reply-To: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> References: <30899801.19161.1241540779042.JavaMail.douglasbecker@127.0.0.1> Message-ID: <4A0012880200007200033005@health-mail2.tpchd.org> Fewer viruses, for one thing. (grin) Seriously, Windows isn't as bad as it used to be. From my perspective it's an inferior but safe bet, in that absent any thought or information management will support a move to all microsoft. Vulnerabilities, cost, and greater management efficiencies are all things that linux CAN bring to the table, but of course the details of implementation always matter. Randy Grein Sr. Network Engineer >>> Douglas Becker 5/5/2009 9:26 AM >>> I agree with you there. My situation is that our accounting vendor stopped supporting and developing for Netware. My reaction was to find another accounting package that did. Unfortunately, the CFO "WILL NOT" (her words) do a migration of the accounting data to another package. Therefore I was stuck with having to get another server (hardware and all) just to put the accounting package on it. The same thing happened again with the fixed asset package. Now I have a server that does nothing but file and print sharing using Netware, one Netware server doing GW and two Windows server for the other stuff. I am using GW 6.5 and to get the pipe to allow the use of Outlook, I need to upgrade to 8, and they won't do that because there is a chance we might just move everything to Windows. I have some say in that, but unfortunately it does not carry a lot of weight. Aside from allowing me to keep GW, what would moving to Linux do for me? Douglas Becker douglasbecker at cox.net On Mon, May 4, 2009 at 12:18 PM , Randy Grein wrote: > Not knowing your requirement it's hard to say. It is common for a > particular application vendor to deploy only on one platform but there > are almost always substitutes. The quality of the substitute may be > unacceptable for the situation. There may be legal requirements, or > political (the owner will accept ONLY Autocad or ONLY Microsoft > Office, not trubocad or Open Office). The amount of work required to > satisfy such a requirement may be too much (running all applications > on linux workstations in a virtual environment is a good example - > possible but inefficient). My point wasn't to say that you missed > something obvious, just that we tend to get locked up in a > one-solution mindset, especially when vendor lockin happens. It's why > economists do not talk about needs, only wants. Even air will be > bartered if something else more valuable is needed. > > > > > > Randy Grein > Sr. Network Engineer >>>> Douglas Becker 5/4/2009 1:07 PM >>> > But I think that you are missing the point with what he was saying. > He said that some of his apps were windows only. I have the same > problem. I cannot get that program in a Linux version. Therefore I am > forced to switch to windows server. And it may be true that the cost > to upgrade to Linux is less, but if it will not run the programs that > I have to run, why would I use it at all? > > Douglas Becker > > douglasbecker at cox.net > > On Mon, May 4, 2009 at 4:03 PM , Randy Grein wrote: > >> Absolutely right Bob. A good-sized company here in Tacoma was >> featured in Computerworld last year; their 3 year Microsoft licensing >> costs paid for a full migration to Apple - workstations and servers, >> with, IIRC $1.2m to spare. The glue used by apple is OpenLDAP which >> is (of course) available on many other platforms. Point is, the cost >> savings are there for many, if not most organizations. >> >> >> >> >> Randy Grein >> Sr. Network Engineer >>>>> Bob Jonkman 5/4/2009 11:38 AM >>> >> I'll bet that the bulk of your computing needs are for standard >> office productivity: Payroll, HR, CRM, and the standard desktop >> applications of word processor, spreadsheet, e-mail and maybe Web >> browsing. I'll bet that you can meet all your computing needs with >> Linux-based software. Yes, there is some cost to switching >> platforms, but I'll also bet that the cost to switch to Linux today >> is less than the cost of upgrading to Microsoft licenses, not to >> mention the Microsoft licence renewal costs year after year. >> >> --Bob. >> >> >> >> ************************************************************************************* >> This e-mail and any attachments may contain confidential and >> privileged information. It has been scanned for viruses. If you are >> not the intended recipient, please notify the sender immediately by >> return e-mail, delete this e-mail and destroy any copies. Any >> dissemination, use, review, disclosure, or distribution of this >> information by a person other than the intended recipient is >> unauthorized and may be illegal. >> >> >> ************************************************************************************** >> >> >> _______________________________________________ >> Novell mailing list >> Novell at netlab1.oucs.ox.ac.uk >> http://netlab1.usu.edu/mailman/listinfo/novell > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > > ************************************************************************************* > This e-mail and any attachments may contain confidential and > privileged information. It has been scanned for viruses. If you are > not the intended recipient, please notify the sender immediately by > return e-mail, delete this e-mail and destroy any copies. Any > dissemination, use, review, disclosure, or distribution of this > information by a person other than the intended recipient is > unauthorized and may be illegal. > > ************************************************************************************** > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From joea at j4computers.com Wed May 6 14:19:32 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Wed, 06 May 2009 09:19:32 -0400 Subject: Linux Swap "preallocate" or not? Message-ID: <4A015611020000850005FC76@FS-LIN-OES> > Joe, > That does seem odd to me. One can create new swap files on the > fly and use swapon as you have. > Normally we create a swap partition of reasonable size and see if it is > consumed. By then, as you also note, > the system ought to become sluggish. I create 256MB or 512MB swap > partitions, depending upon expected > system usage, and they rarely consume that much space. The system does not seem sluggish, at all, even when "swap" appears almost entirely consumed. I have been looking for a means to track when (if) swap is actually being written to, without much success. > So, first, what does "top" say about swap usage? Then is there > something on that system which > keeps asking for more and more memory? >We can use the "ps aux" command to see both real and > virtual memory usage. Some turning off of services one by one might > reveal the culprit. . . . We can see, from top, that it is the gwpoa (groupwise post office agents) that are the major consumers. Also, on another box, a GW MTA began playing the same game. We restarted that and it immediately gave up the swap and has been nice since. This seems odd behavior, but it does coincide with two things we began doing. One was a rescehduling, for more frequent "quickfinder indexing", which was the orignal suspect. But changing that back to "normal" and watching trends shows that did not change matters. We also began doing manual "eDirectory sync" from a GW perspective, for all users in the system. (reads edir, refreshes the info in GroupWise that corresponds). That does seem to affect trends. We see "a lot" of MTA and POA activity when that is in progress and swap usage does seem to increase. Fine. Problem is, it never appears to give it up. joe a. From joea at j4computers.com Wed May 6 14:40:41 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Wed, 06 May 2009 09:40:41 -0400 Subject: Linux Swap "preallocate" or not? Message-ID: <4A015B17020000850005FC84@FS-LIN-OES> >We can see, from top, that it is the gwpoa (groupwise post office agents) that are the major consumers. >Also, on another box, a GW MTA began playing the same game. We restarted that and it immediately gave up >the swap and has been nice since. >This seems odd behavior, I mean the overall behavior, it is not odd at all, to me, that it should give up the swap, on restart. Sorry for not being clear. joe a. From joe.doupnik at oucs.ox.ac.uk Wed May 6 15:06:06 2009 From: joe.doupnik at oucs.ox.ac.uk (Joe Doupnik) Date: Wed, 06 May 2009 15:06:06 +0100 Subject: Linux Swap "preallocate" or not? In-Reply-To: <4A015611020000850005FC76@FS-LIN-OES> References: <4A015611020000850005FC76@FS-LIN-OES> Message-ID: <4A01994E.2070705@oucs.ox.ac.uk> joea at j4computers.com wrote: >> Joe, >> That does seem odd to me. One can create new swap files on the >> fly and use swapon as you have. >> Normally we create a swap partition of reasonable size and see if it is >> consumed. By then, as you also note, >> the system ought to become sluggish. I create 256MB or 512MB swap >> partitions, depending upon expected >> system usage, and they rarely consume that much space. > > The system does not seem sluggish, at all, even when "swap" appears almost entirely consumed. > I have been looking for a means to track when (if) swap is actually being written to, without much success. > >> So, first, what does "top" say about swap usage? Then is there >> something on that system which >> keeps asking for more and more memory? >> We can use the "ps aux" command to see both real and >> virtual memory usage. Some turning off of services one by one might >> reveal the culprit. . . . > > We can see, from top, that it is the gwpoa (groupwise post office agents) that are the major consumers. > Also, on another box, a GW MTA began playing the same game. We restarted that and it immediately gave up > the swap and has been nice since. > > This seems odd behavior, but it does coincide with two things we began doing. One was a rescehduling, for more > frequent "quickfinder indexing", which was the orignal suspect. But changing that back to "normal" and watching trends > shows that did not change matters. > > We also began doing manual "eDirectory sync" from a GW perspective, for all users in the > system. (reads edir, refreshes the info in GroupWise that corresponds). That does seem to affect trends. We see "a lot" of MTA and POA activity when that is in progress and swap usage does seem to increase. Fine. Problem is, it never appears to give it up. > > joe a. ------------- Java has the characteristic of using lots of virtual (and physical) memory, and it is supposed to gradually release allocations gradually over tens of minutes. That leaves the applications themselves to cooperate by free-ing old allocations. Quickfinder can consume much memory while indexing, but it tries to be kind and not overdo allocation. It is good this way. Very very large archives can take "awhile" to process and have the partial indices merged into the final one. For what it is worth dept, I did some GW stress testing this past winter by feeding it about half a million messages (incoming) as quickly as it could accept them. I did not see unusual swap file usage; GW was well behaved on my gear. Note that this was on a 64-bit machine with matching SLES. Matters become more stressful on 32-bit machines. You might check the eDir cache memory usage and hit ratios. Recall, dynamic cache allocation can allow very large temporary consumption, so consider hard limited. Joe D. From marty at toombs.com Thu May 7 02:41:12 2009 From: marty at toombs.com (Marty Toombs) Date: Wed, 06 May 2009 21:41:12 -0400 Subject: Whither Novell In-Reply-To: References: Message-ID: <4A023C38.50309@toombs.com> Just to respond to some follow-up to my earlier comments regarding being forced to move to Windows by Novell: 1. Some writers had criticized comparing Windows Server 2008 to Netware 5.1 because of the age differences. That's why I noted I'd love to use Netware 7 and do that comparison, but Novell decided not to go that way. 2. Application requirements. We are a fairly small ($27 million) organization so I'm aware of the immense cost involved in a change in general ledger or payroll software. The last time we changed payroll vendors I was the one burning the midnight oil in Christmas week to move the data. When those vendors stopped supporting Netware, requiring a move to Windows, I had no way to argue that we should change vendors just to avoid Windows. We needed a new medical billing package that runs only on Windows. So the business office has one Netware server for most of its work, and then 3 Windows boxes, one for payroll, one for general ledger and one for medical billing. A fourth Windows server provides terminal server access. Our servers outside the business office remain Netware for now. 3. We are heavily dependent on Microsoft Access applications. About three-quarters of our revenue comes from work we record, document and bill from Access applications. While that doesn't mean we have to run Windows servers, it does mean we need to use Microsoft Office to keep the lights turned on. 4. Licensing issues mean different things to me than they mean to those in the for-profit world, I realize that. My 600 Netware 6.5 (some upgrade, some new) licenses cost $485. Licensing cost is just one factor for me. (I can get Windows Server 2008 Standard for $110 (no software assurance). Even at that I have much more invested in Windows licenses than Novell licenses even before we do the conversion to Windows. 5. As the agency's HIPAA Security Officer I get to decide in some areas what is secure and what isn't (I'd just better not make a wrong decision ...). The problem is that the auditors checking on me all want to know how my domain controller is secured, etc. When I tell them I don't have one, there is a long pause. Like it or not, they frame all the security questions in a Windows context and it's a battle to force them to think of it in any other way. Securing a Windows box can be a pain, but I'm not sure I could ever convince them that a Linux server had sufficient security, user activity logging, etc., that they want to see. 6. Regarding support, we do our best to avoid having to hire outside contractors. We've hired three staff in the past 18 months. The candidates we interviewed had varying levels of Windows Server experience, but none of them had any experience with Linux servers. On Fri, May 1, 2009 at 4:48 PM , Matt Smith wrote: > > , Randy Grein > > wrote: >> >> Sorry Marty, but I can't buy the logic you're using. Can you explain> >> >> Linux has steadily increased in market share, eating mostly into >> >> Netware (but, AFAICT Windows as well) so that it's somewhere between>> 25-35% of new installs. Support is available from multiple vendors >> >> (open source, SUSE, Red Hat, Debian, Ubuntu, etc) so what's the real>> problem? >> >> >> >> Keep in mind the distinction between security and profit. Security >> >> calls for following the herd. Profit calls for striking out on you >> >> own (correct) path. How much do you trust your own analysis? > > > > I don't want to "gang up" on you, Marty, but my first response was > > pretty much the same as Randy's here. We do have several windows > > servers running as application servers. However, I've found lately > > that > > we were installing Windows to run these application not because they > > were Windows only, but because the sales people pitched it as a > > Windows > > application to the power-that-be. Turns out that if we'd asked, we > > could have been running some of this stuff on Linux after all. I'm > > particularly burned by the recent purchase of our HR system. The > > company that owns the product actually develops on Linux and ports it > > over to Windows! > > > > I'm not being ingenuous. Windows dominates the market. Still, my > > boss > > loves to brag when he hears the latest school system that lost a day > > or > > two's work to the latest internet nasty while we soldier on without a > > hitch. > > > > Besides, SUSE is linux. I figure if I have to, we'll just switch to a > > different flavor. I don't expect my experience to be quite as rich as > > what I get with Novell, but I also don't feel like I'm wasting my time > > now by any means. > > > > Each to their own though. > > > > -Matt > > -- > > > > Matt Smith Network Technology Specialist > > Oconee County School System, Oconee County, Georgia > > Office of Instruction and Technology 706-769-5685 x1314 > > _______________________________________________ > > Novell mailing list > > Novell at netlab1.oucs.ox.ac.uk > > http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Thu May 7 16:01:45 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 07 May 2009 11:01:45 -0400 Subject: Oplocks, ncpserver, ConsoleOne, linux Message-ID: <4A02BF92020000850005FCB5@FS-LIN-OES> Does OpLocks setting have any meaning for non Novell Client applications? OES1 Linux, using ncpserver (ncpcon) to publish ncp volumes, in order to mount them on an "admin" server, where we run ConsoleOne. Having had some interesting issues with apparent db corruption (GW domains), revisited oplocks. TID's etc, seem to focus on the Novell Client (windows) as being involved. Memory holds fragments of previous discussions that seem to say this is not an issue if not using Windows/Novell client. But it strikes me that ConsoleOne may attempt similar shenanigans with file locking, etc. I followed TID 7001108 and used Remote Manager to set oplocks to 0. Even saw it was written to /etc/opt/novell/ncpserv.conf. However, did not see ndsd restart, the manual change section claims must be done. I would have thought it would really require the volumes to be dismounted, at both ends, for the change to take effect, and would have little to do with eDirectory. Illumination? joe a. From joea at j4computers.com Thu May 7 17:10:33 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 07 May 2009 12:10:33 -0400 Subject: Oplocks, ncpserver, ConsoleOne, linux Message-ID: <4A02CFB8020000850005FCB9@FS-LIN-OES> Never Mind. TID 7002459. Goodness me. joe a. >>> "joea at j4computers.com" 05/07/09 11:04 AM >>> Does OpLocks setting have any meaning for non Novell Client applications? OES1 Linux, using ncpserver (ncpcon) to publish ncp volumes, in order to mount them on an "admin" server, where we run ConsoleOne. Having had some interesting issues with apparent db corruption (GW domains), revisited oplocks. TID's etc, seem to focus on the Novell Client (windows) as being involved. Memory holds fragments of previous discussions that seem to say this is not an issue if not using Windows/Novell client. But it strikes me that ConsoleOne may attempt similar shenanigans with file locking, etc. I followed TID 7001108 and used Remote Manager to set oplocks to 0. Even saw it was written to /etc/opt/novell/ncpserv.conf. However, did not see ndsd restart, the manual change section claims must be done. I would have thought it would really require the volumes to be dismounted, at both ends, for the change to take effect, and would have little to do with eDirectory. Illumination? joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From TJohnson at lancaster.wnyric.org Fri May 8 15:52:35 2009 From: TJohnson at lancaster.wnyric.org (TJohnson at lancaster.wnyric.org) Date: Fri, 8 May 2009 10:52:35 -0400 Subject: Persistent device names for iSCSI disks Message-ID: I posted this over on the TTP list but I figured someone may have a clue for me on this one: I am setting up a new SLES10SP2/OES2SP1 cluster using iSCSI shared storage and NCS. i setup my iSCSI disk and set them to login automatically. My SBD LUN on the iSCSI SAN showed up as /dev/sda and my other iSCSI LUN is /dev/sdb on both boxes. I applied some updates on the servers, rebooted go to configure NCS and now I see the SBD LUN on 1 server is /dev/sdb and on the other server it is /dev/sda, I reboot first server and the SBD LUN is back on /dev/sda I have never seen this happen with any of my other iSCSI LUNs but it has me thinking if it will happen again and what kind of havoc will it wreak on my cluster. So my question, is there a way to "pin" a device name (/dev/sda, /dev/sdb, etc.) to a certain LUN and/or is there a way to use /dev/by-id for iSCSI? I have done some searches ban found some info on using udev to accomplish this but I figured i would run it by the list. Thanks. T2 From budthegrey at gmail.com Fri May 8 20:00:15 2009 From: budthegrey at gmail.com (Bud Durland) Date: Fri, 8 May 2009 15:00:15 -0400 Subject: eDir sync behind a firewall Message-ID: <509526700905081200h7892ba9ehed4373f2a80f11a2@mail.gmail.com> A friend called with a problem. He's adding a new NetWare 6.5/sp8 server to his tree. The other servers are also NW6.5/Sp8 The new server is in a building down the road. There is a firewall (SmoothWall) installed at each location, and the NetWare servers are behind them. He's having trouble with NDS synchronization. He tells me that all the appropriate ports are opened on the firewalls. If he uses DSREPAIR and requests and "Unattended Full Repair", it all goes well. But, a while later NDS is out of sync between the two. He tells me that TimeSync is working Ok and does not go out of sync. I know he's using NTP, but I don't know how it's configured. I'm wondering if the problem is that because the servers are behind firewalls, they report their internal IP address (192.168.x.x), but the packets appear to come from the external address (152.153.x.x). I did a cursory search of the knowledge base, but didn't see anything that looked like it would help here. Any help appreciated. -- ---------------------------------------------------------- I'm in my own little world. But that's OK, because they all know me here. From James.Taylor at eastcobbgroup.com Fri May 8 20:04:36 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Fri, 08 May 2009 15:04:36 -0400 Subject: eDir sync behind a firewall In-Reply-To: <509526700905081200h7892ba9ehed4373f2a80f11a2@mail.gmail.com> References: <509526700905081200h7892ba9ehed4373f2a80f11a2@mail.gmail.com> Message-ID: <4A044A040200007500032357@inet.eastcobbgroup.com> eDir will not sync across a NAT. They will need to set up a VPN to make this work. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> Bud Durland 5/8/2009 03:00 PM >>> A friend called with a problem. He's adding a new NetWare 6.5/sp8 server to his tree. The other servers are also NW6.5/Sp8 The new server is in a building down the road. There is a firewall (SmoothWall) installed at each location, and the NetWare servers are behind them. He's having trouble with NDS synchronization. He tells me that all the appropriate ports are opened on the firewalls. If he uses DSREPAIR and requests and "Unattended Full Repair", it all goes well. But, a while later NDS is out of sync between the two. He tells me that TimeSync is working Ok and does not go out of sync. I know he's using NTP, but I don't know how it's configured. I'm wondering if the problem is that because the servers are behind firewalls, they report their internal IP address (192.168.x.x), but the packets appear to come from the external address (152.153.x.x). I did a cursory search of the knowledge base, but didn't see anything that looked like it would help here. Any help appreciated. -- ---------------------------------------------------------- I'm in my own little world. But that's OK, because they all know me here. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From mbrady at ingenuityieq.com Fri May 8 20:06:29 2009 From: mbrady at ingenuityieq.com (Mike Brady) Date: Fri, 08 May 2009 15:06:29 -0400 Subject: eDir sync behind a firewall In-Reply-To: <509526700905081200h7892ba9ehed4373f2a80f11a2@mail.gmail.com> References: <509526700905081200h7892ba9ehed4373f2a80f11a2@mail.gmail.com> Message-ID: <4A044A750200002D000144D5@mail.ingenuityieq.com> Could be a problem with SLP propagation (or lack of)? >>> On 5/8/2009 at 3:00 PM, in message <509526700905081200h7892ba9ehed4373f2a80f11a2 at mail.gmail.com>, Bud Durland wrote: A friend called with a problem. He's adding a new NetWare 6.5/sp8 server to his tree. The other servers are also NW6.5/Sp8 The new server is in a building down the road. There is a firewall (SmoothWall) installed at each location, and the NetWare servers are behind them. He's having trouble with NDS synchronization. He tells me that all the appropriate ports are opened on the firewalls. If he uses DSREPAIR and requests and "Unattended Full Repair", it all goes well. But, a while later NDS is out of sync between the two. He tells me that TimeSync is working Ok and does not go out of sync. I know he's using NTP, but I don't know how it's configured. I'm wondering if the problem is that because the servers are behind firewalls, they report their internal IP address (192.168.x.x), but the packets appear to come from the external address (152.153.x.x). I did a cursory search of the knowledge base, but didn't see anything that looked like it would help here. Any help appreciated. -- ---------------------------------------------------------- I'm in my own little world. But that's OK, because they all know me here. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Fri May 8 20:11:43 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 08 May 2009 15:11:43 -0400 Subject: eDir sync behind a firewall Message-ID: <4A044BA8020000850005FCD7@FS-LIN-OES> Is this over the Internet? I suspect he will need to setup a VPN between the two locations, as I think (have heard) eDirectory will never be happy otherwise. joe a. >>> Bud Durland 05/08/09 3:02 PM >>> A friend called with a problem. He's adding a new NetWare 6.5/sp8 server to his tree. The other servers are also NW6.5/Sp8 The new server is in a building down the road. There is a firewall (SmoothWall) installed at each location, and the NetWare servers are behind them. He's having trouble with NDS synchronization. He tells me that all the appropriate ports are opened on the firewalls. If he uses DSREPAIR and requests and "Unattended Full Repair", it all goes well. But, a while later NDS is out of sync between the two. He tells me that TimeSync is working Ok and does not go out of sync. I know he's using NTP, but I don't know how it's configured. I'm wondering if the problem is that because the servers are behind firewalls, they report their internal IP address (192.168.x.x), but the packets appear to come from the external address (152.153.x.x). I did a cursory search of the knowledge base, but didn't see anything that looked like it would help here. Any help appreciated. -- ---------------------------------------------------------- I'm in my own little world. But that's OK, because they all know me here. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Fri May 8 20:36:09 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 08 May 2009 15:36:09 -0400 Subject: OES1 Linux ncp volumes, eDirectory Message-ID: <4A045164020000850005FCE3@FS-LIN-OES> Newly created ncp(con) volumes on a linux, oes1 box, to not show up in eDirectory. In some cases, created new, from scratch, in others deleted old volumes names and created new volume names. Neither do those show up, in eDirectory. Did rcndsd restart, rcslpd restart for giggles. Can see them doing browse, via windows network neighborhood/netware services. I can access the files. Simply cannot get them to show up in eDirectory. I have a dim recollection, but nothing from memory or KB searches. If there is a Netware like facility to "upgrade mounted volumes . . ." I do not find it. joe a. From petervl at gmail.com Mon May 11 18:25:41 2009 From: petervl at gmail.com (Peter Van Lone) Date: Mon, 11 May 2009 12:25:41 -0500 Subject: licensing NW65 Message-ID: <68b791330905111025wbe251f1ge7d3c4ffa72ea9e4@mail.gmail.com> I believe that all licensing "hard stops" have been removed from Netware at this point, is that correct? So, I've got a company that wants to install a new NW65 server as an upgrade to an existing NW6.0 server .. they have all of their users licensed -- do we need to worry about finding a license file to install with the new server? (they already have many nw65 servers .. this is just one server in one of their smaller locations) I know that a license file for server encryption is on the install cd ... so we can browse for that during install ... I just want confirmation that I do not need to install a NW65 "user license" .... Also -- they will be installing from a NW65 sp6 over-lay because that is what they have in the tree at this point. Peter ------------------------------- "Curiosity is one of the most permanent and certain characteristics of a vigorous intellect." ~Samuel Johnson http://www.the-brights.net http://xkcd.com/167 From RGrein at tpchd.org Mon May 11 18:27:17 2009 From: RGrein at tpchd.org (Randy Grein) Date: Mon, 11 May 2009 10:27:17 -0700 Subject: licensing NW65 In-Reply-To: <68b791330905111025wbe251f1ge7d3c4ffa72ea9e4@mail.gmail.com> References: <68b791330905111025wbe251f1ge7d3c4ffa72ea9e4@mail.gmail.com> Message-ID: <4A07FD850200007200033536@health-mail2.tpchd.org> Correct on all counts. Randy Grein Sr. Network Engineer >>> Peter Van Lone 5/11/2009 10:25 AM >>> I believe that all licensing "hard stops" have been removed from Netware at this point, is that correct? So, I've got a company that wants to install a new NW65 server as an upgrade to an existing NW6.0 server .. they have all of their users licensed -- do we need to worry about finding a license file to install with the new server? (they already have many nw65 servers .. this is just one server in one of their smaller locations) I know that a license file for server encryption is on the install cd ... so we can browse for that during install ... I just want confirmation that I do not need to install a NW65 "user license" .... Also -- they will be installing from a NW65 sp6 over-lay because that is what they have in the tree at this point. Peter ------------------------------- "Curiosity is one of the most permanent and certain characteristics of a vigorous intellect." ~Samuel Johnson http://www.the-brights.net http://xkcd.com/167 _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From joea at j4computers.com Mon May 11 20:06:38 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Mon, 11 May 2009 15:06:38 -0400 Subject: OES1 Linux ncp volumes, eDirectory Message-ID: <4A083EF7020000850005FD0B@FS-LIN-OES> Seems one must create these using ncpcon to have them show up in eDirectory. And even, in some cases, to have them show up via Windows browse. Attempting to create them via manual entries in /etc/ncpserv.conf is problematic. An interesing note is that when creating them with ncpcon and adding "Inherit Posix Permissions" to the command line, that does not get added to /etc/ncpserv.conf. The volume name and path get created, but Inherit Posix Permissions does not. joe a. >>> "joea at j4computers.com" 05/08/09 3:38 PM >>> Newly created ncp(con) volumes on a linux, oes1 box, to not show up in eDirectory. In some cases, created new, from scratch, in others deleted old volumes names and created new volume names. Neither do those show up, in eDirectory. Did rcndsd restart, rcslpd restart for giggles. Can see them doing browse, via windows network neighborhood/netware services. I can access the files. Simply cannot get them to show up in eDirectory. I have a dim recollection, but nothing from memory or KB searches. If there is a Netware like facility to "upgrade mounted volumes . . ." I do not find it. joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From craighewitt at massan.co.uk Tue May 12 12:44:42 2009 From: craighewitt at massan.co.uk (craighewitt at massan.co.uk) Date: Tue, 12 May 2009 12:44:42 +0100 (BST) Subject: NSS Utils In-Reply-To: <4A083EF7020000850005FD0B@FS-LIN-OES> References: <4A083EF7020000850005FD0B@FS-LIN-OES> Message-ID: <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> Hi, Does anyone know of any utils out there that can report what files are owned by what users. We run OES2 Linux and implement quotas so we'd like to be able to see what files are eating their quota and also would be good if non admin users could run it too. Thanks, Craig. From mrsmith at oconee.k12.ga.us Tue May 12 13:16:38 2009 From: mrsmith at oconee.k12.ga.us (Matt Smith) Date: Tue, 12 May 2009 08:16:38 -0400 Subject: NSS Utils In-Reply-To: <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> Message-ID: <4A093051.E4C1.0068.0@oconee.k12.ga.us> >>> On 5/12/2009 at 7:44 AM, in message <53899.81.149.45.143.1242128682.squirrel at www.massan.co.uk>, wrote: > Hi, > > Does anyone know of any utils out there that can report what files are > owned by what users. We run OES2 Linux and implement quotas so we'd like > to be able to see what files are eating their quota and also would be good > if non admin users could run it too. > > Thanks, > > Craig. You could always check out the file/volume reporting in NRM, https::8009. It actually has comprehensive reporting, and should answer your questions. I haven't found an easy way to easily export the NRM reports to a file though. So far I've just copied and pasted into a spreadsheet if I needed something more than what NRM shows. You do want to be a little careful when you tell NRM to run a report though. I have seen some of the more intensive reports really impact a server's performance while they were running. -Matt -- Matt Smith Network Technology Specialist Oconee County School System, Oconee County, Georgia Office of Instruction and Technology 706-769-5685 x1314 From jgramse at utah.gov Tue May 12 13:39:47 2009 From: jgramse at utah.gov (Jim Gramse) Date: Tue, 12 May 2009 06:39:47 -0600 Subject: NSS Utils In-Reply-To: <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> Message-ID: <4A0919B3020000760000B0B4@gwia2.state.ut.us> The Novell remote manager screen http://serveraddress:8008 then choose volume then thefolder and hit inventorythe next screen will shoe files and size and a whole lot more as long as you are using NSS volumes. >>> 5/12/2009 5:44 AM >>> Hi, Does anyone know of any utils out there that can report what files are owned by what users. We run OES2 Linux and implement quotas so we'd like to be able to see what files are eating their quota and also would be good if non admin users could run it too. Thanks, Craig. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From alandpearson at yahoo.com Tue May 12 14:25:07 2009 From: alandpearson at yahoo.com (Alan Pearson) Date: Tue, 12 May 2009 14:25:07 +0100 (BST) Subject: NSS Utils In-Reply-To: <4A0919B3020000760000B0B4@gwia2.state.ut.us> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> <4A0919B3020000760000B0B4@gwia2.state.ut.us> Message-ID: <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> This isn't really that helpful 1) Running on Linux it shows numeric userids, for users, not the username 2) You can't get a report per user with a breakdown of what files are in use Something that the users could access themselves would be the way forward, where they could see what files they own, and the size each takes. Nearest I've seen is a tool called Disk Space Usage http://www.simonsware.com/downloads.html but it's not ideal and is slooow -- AlanP On Tue, May 12, 2009 1:39 pm, Jim Gramse wrote: > The Novell remote manager screen http://serveraddress:8008 then choose > volume then thefolder and hit inventorythe next screen will shoe files and > size and a whole lot more as long as you are using NSS volumes. > >>>> 5/12/2009 5:44 AM >>> > Hi, > > Does anyone know of any utils out there that can report what files are > owned by what users. We run OES2 Linux and implement quotas so we'd like > to be able to see what files are eating their quota and also would be good > if non admin users could run it too. > > Thanks, > > Craig. > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > From mrsmith at oconee.k12.ga.us Tue May 12 15:08:20 2009 From: mrsmith at oconee.k12.ga.us (Matt Smith) Date: Tue, 12 May 2009 10:08:20 -0400 Subject: NSS Utils In-Reply-To: <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> <4A0919B3020000760000B0B4@gwia2.state.ut.us> <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> Message-ID: <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> On 5/12/2009 at 9:25 AM, in message <46084.88.211.54.85.1242134707.squirrel at 83.67.10.8>, "Alan Pearson" wrote: > This isn't really that helpful > > 1) Running on Linux it shows numeric userids, for users, not the username > 2) You can't get a report per user with a breakdown of what files are in use > > Something that the users could access themselves would be the way forward, > where they could see what files they own, and the size each takes. > > Nearest I've seen is a tool called Disk Space Usage > http://www.simonsware.com/downloads.html > > but it's not ideal and is slooow You know, you're right. It really isn't as useful on OES Linux. I wonder if Novell plans on "fixing" that? Perhaps there's something in the mighty JRB utilities that would suffice? I'd look but need to concentrate elsewhere right now. I am interested in seeing where this threat leads though. -Matt -- Matt Smith Network Technology Specialist Oconee County School System, Oconee County, Georgia Office of Instruction and Technology 706-769-5685 x1314 From joea at j4computers.com Tue May 12 15:43:21 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 12 May 2009 10:43:21 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0952BB020000850005FD25@FS-LIN-OES> It is said that, in NetWare Clustering, one could map to a volume contained in a cluster object, without reference to the hosting node. That feature does not seem to be available in the Linux flavor of NCS. Or am I missing something? I can map to the volume, command line or login script, if I reference the the server and volume. But, attempting to map to the "Cluster Object" always fails. No NetWare cluster here to compare to. joe a. From RGrein at tpchd.org Tue May 12 15:45:48 2009 From: RGrein at tpchd.org (Randy Grein) Date: Tue, 12 May 2009 07:45:48 -0700 Subject: NSS Utils In-Reply-To: <4A093051.E4C1.0068.0@oconee.k12.ga.us> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> <4A093051.E4C1.0068.0@oconee.k12.ga.us> Message-ID: <4A09292C020000720003365D@health-mail2.tpchd.org> Those reports are indeed disk intensive. No major hit on processor or memory; a good way to tell if you MIGHT need more drives. (grin) Randy Grein Sr. Network Engineer >>> "Matt Smith" 5/12/2009 5:16 AM >>> >>> On 5/12/2009 at 7:44 AM, in message <53899.81.149.45.143.1242128682.squirrel at www.massan.co.uk>, wrote: > Hi, > > Does anyone know of any utils out there that can report what files are > owned by what users. We run OES2 Linux and implement quotas so we'd like > to be able to see what files are eating their quota and also would be good > if non admin users could run it too. > > Thanks, > > Craig. You could always check out the file/volume reporting in NRM, https::8009. It actually has comprehensive reporting, and should answer your questions. I haven't found an easy way to easily export the NRM reports to a file though. So far I've just copied and pasted into a spreadsheet if I needed something more than what NRM shows. You do want to be a little careful when you tell NRM to run a report though. I have seen some of the more intensive reports really impact a server's performance while they were running. -Matt -- Matt Smith Network Technology Specialist Oconee County School System, Oconee County, Georgia Office of Instruction and Technology 706-769-5685 x1314 _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From tim at nds8.co.uk Tue May 12 15:48:00 2009 From: tim at nds8.co.uk (Tim Heywood) Date: Tue, 12 May 2009 15:48:00 +0100 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0952BB020000850005FD25@FS-LIN-OES> References: <4A0952BB020000850005FD25@FS-LIN-OES> Message-ID: <4A099A30020000BB0005B824@mail2.nds8.com> Every NSS resource has a virtual Server object created automatically so that (like in NetWare) you can map to the resource and not the IP etc. However, the posix based resources do not have the virtual server created automatically... It was supposed to be an improvement for SP1 - indeed I was told it was included - but it seems I was misinformed (or should that be that someone "misspoke?") and you still have to create the Virtual server manually. T -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 15:43, "joea at j4computers.com" wrote: > It is said that, in NetWare Clustering, one could map to a volume contained > in a cluster object, without reference to the hosting node. That feature > does not seem to be available in the Linux flavor of NCS. Or am I missing > something? > > I can map to the volume, command line or login script, if I reference the > the server and volume. But, attempting to map to the "Cluster Object" always > fails. > > No NetWare cluster here to compare to. > > joe a. > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > ***Scanned by M+ Guardian*** The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. From RGrein at tpchd.org Tue May 12 15:48:05 2009 From: RGrein at tpchd.org (Randy Grein) Date: Tue, 12 May 2009 07:48:05 -0700 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0952BB020000850005FD25@FS-LIN-OES> References: <4A0952BB020000850005FD25@FS-LIN-OES> Message-ID: <4A0929B50200007200033661@health-mail2.tpchd.org> Works that way with Netware cluster - been doing it for years. Randy Grein Sr. Network Engineer >>> "joea at j4computers.com" 5/12/2009 7:43 AM >>> It is said that, in NetWare Clustering, one could map to a volume contained in a cluster object, without reference to the hosting node. That feature does not seem to be available in the Linux flavor of NCS. Or am I missing something? I can map to the volume, command line or login script, if I reference the the server and volume. But, attempting to map to the "Cluster Object" always fails. No NetWare cluster here to compare to. joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ************************************************************************************* This e-mail and any attachments may contain confidential and privileged information. It has been scanned for viruses. If you are not the intended recipient, please notify the sender immediately by return e-mail, delete this e-mail and destroy any copies. Any dissemination, use, review, disclosure, or distribution of this information by a person other than the intended recipient is unauthorized and may be illegal. ************************************************************************************** From joea at j4computers.com Tue May 12 16:22:41 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 12 May 2009 11:22:41 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A095BF5020000850005FD2C@FS-LIN-OES> These are "non NSS" volumes. Any pointers on how to do this "manual" create? I'll first have to hunt down C1 plugins, for NCS, no doubt, as some items in the cluster area have the "?" icon. joe a. >>> "Tim Heywood" 05/12/09 10:49 AM >>> Every NSS resource has a virtual Server object created automatically so that (like in NetWare) you can map to the resource and not the IP etc. However, the posix based resources do not have the virtual server created automatically... It was supposed to be an improvement for SP1 - indeed I was told it was included - but it seems I was misinformed (or should that be that someone "misspoke?") and you still have to create the Virtual server manually. T -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 15:43, "joea at j4computers.com" wrote: > It is said that, in NetWare Clustering, one could map to a volume contained > in a cluster object, without reference to the hosting node. That feature > does not seem to be available in the Linux flavor of NCS. Or am I missing > something? > > I can map to the volume, command line or login script, if I reference the > the server and volume. But, attempting to map to the "Cluster Object" always > fails. > > No NetWare cluster here to compare to. > > joe a. > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > ***Scanned by M+ Guardian*** The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From tim at nds8.co.uk Tue May 12 16:39:02 2009 From: tim at nds8.co.uk (Tim Heywood) Date: Tue, 12 May 2009 16:39:02 +0100 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A095BF5020000850005FD2C@FS-LIN-OES> References: <4A095BF5020000850005FD2C@FS-LIN-OES> Message-ID: <4A09A626020000BB0005B880@mail2.nds8.com> You got it - I don't have a cheat sheet to hand but the documentation is not bad... http://www.novell.com/documentation/oes2/file_ncp_lx/index.html?page=/documentation/oes2/file_ncp_lx/data/bceq9za.html -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 16:22, "joea at j4computers.com" wrote: > These are "non NSS" volumes. > > Any pointers on how to do this "manual" create? I'll first have to hunt > down C1 plugins, for NCS, no doubt, as some items in the cluster area have > the "?" icon. > > joe a. > >>>> "Tim Heywood" 05/12/09 10:49 AM >>> > Every NSS resource has a virtual Server object created automatically so > that (like in NetWare) you can map to the resource and not the IP etc. > However, the posix based resources do not have the virtual server > created automatically... It was supposed to be an improvement for SP1 - > indeed I was told it was included - but it seems I was misinformed (or > should that be that someone "misspoke?") and you still have to create > the Virtual server manually. > > T The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. From joea at j4computers.com Tue May 12 16:56:20 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 12 May 2009 11:56:20 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0963DF020000850005FD30@FS-LIN-OES> I'll check out the doc. However, this is OES1. No go? joe a. >>> "Tim Heywood" 05/12/09 11:41 AM >>> You got it - I don't have a cheat sheet to hand but the documentation is not bad... http://www.novell.com/documentation/oes2/file_ncp_lx/index.html?page=/documentation/oes2/file_ncp_lx/data/bceq9za.html -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 16:22, "joea at j4computers.com" wrote: > These are "non NSS" volumes. > > Any pointers on how to do this "manual" create? I'll first have to hunt > down C1 plugins, for NCS, no doubt, as some items in the cluster area have > the "?" icon. > > joe a. > >>>> "Tim Heywood" 05/12/09 10:49 AM >>> > Every NSS resource has a virtual Server object created automatically so > that (like in NetWare) you can map to the resource and not the IP etc. > However, the posix based resources do not have the virtual server > created automatically... It was supposed to be an improvement for SP1 - > indeed I was told it was included - but it seems I was misinformed (or > should that be that someone "misspoke?") and you still have to create > the Virtual server manually. > > T The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From tim at nds8.co.uk Tue May 12 17:49:14 2009 From: tim at nds8.co.uk (Tim Heywood) Date: Tue, 12 May 2009 17:49:14 +0100 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0963DF020000850005FD30@FS-LIN-OES> References: <4A0963DF020000850005FD30@FS-LIN-OES> Message-ID: <4A09B69A020000BB0005B8D7@mail2.nds8.com> Can't use the script - but otherwise it still works (or did on the 4 clusters where we have sorted this before) HTH T -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 16:56, "joea at j4computers.com" wrote: > I'll check out the doc. However, this is OES1. No go? > > joe a. > > >>>> "Tim Heywood" 05/12/09 11:41 AM >>> > You got it - I don't have a cheat sheet to hand but the documentation is > not bad... > > http://www.novell.com/documentation/oes2/file_ncp_lx/index.html?page=/docume > ntation/oes2/file_ncp_lx/data/bceq9za.html The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. From joea at j4computers.com Tue May 12 19:00:57 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 12 May 2009 14:00:57 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A098115020000850005FD39@FS-LIN-OES> What about "Inherit_POSIX_Permissions"? Will I be able to specify that using the ncpcon commands in OES1? Need to know this, before going much further, as it will be a show stopper. joe a. >>> "Tim Heywood" 05/12/09 12:55 PM >>> Can't use the script - but otherwise it still works (or did on the 4 clusters where we have sorted this before) HTH T -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 12 May, 2009 at 16:56, "joea at j4computers.com" wrote: > I'll check out the doc. However, this is OES1. No go? > > joe a. > > >>>> "Tim Heywood" 05/12/09 11:41 AM >>> > You got it - I don't have a cheat sheet to hand but the documentation is > not bad... > > http://www.novell.com/documentation/oes2/file_ncp_lx/index.html?page=/docume > ntation/oes2/file_ncp_lx/data/bceq9za.html The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From alandpearson at yahoo.com Tue May 12 19:44:33 2009 From: alandpearson at yahoo.com (Alan Pearson) Date: Tue, 12 May 2009 19:44:33 +0100 Subject: NSS Utils In-Reply-To: <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> <4A0919B3020000760000B0B4@gwia2.state.ut.us> <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> Message-ID: ok, little update NRM, although not suited to the task is useful. The UserID problem goes away if the users are LUM enabled on the server NRM runs on. This is still poor, and IMO, NRM should go to edir for file owners, not local OS. If the users are LUM enabled, you can do some stuff on linux using _real_ tools for the job : find -uid -size +100000 /media/nss/VOLUME will list files > 100Mb on VOLUME owned by uid. With this knowlege you can write scripts using find & the du command to do some basic stuff and produce reports. I wonder if you can get the amount of quota used by a user from _adminfs_ volume ? That would be cool and could be used as a trigger to generate reports when a user hits > than % of quota. Me goes off to investigate.... --- AlanP On 12 May 2009, at 15:08, Matt Smith wrote: > On 5/12/2009 at 9:25 AM, in message > <46084.88.211.54.85.1242134707.squirrel at 83.67.10.8>, "Alan Pearson" > wrote: >> This isn't really that helpful >> >> 1) Running on Linux it shows numeric userids, for users, not the > username >> 2) You can't get a report per user with a breakdown of what files are > in use >> >> Something that the users could access themselves would be the way > forward, >> where they could see what files they own, and the size each takes. >> >> Nearest I've seen is a tool called Disk Space Usage >> http://www.simonsware.com/downloads.html >> >> but it's not ideal and is slooow > > You know, you're right. It really isn't as useful on OES Linux. I > wonder if Novell plans on "fixing" that? > > Perhaps there's something in the mighty JRB utilities that would > suffice? I'd look but need to concentrate elsewhere right now. I am > interested in seeing where this threat leads though. > > -Matt > > -- > > Matt Smith Network Technology Specialist > Oconee County School System, Oconee County, Georgia > Office of Instruction and Technology 706-769-5685 x1314 > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From alandpearson at yahoo.com Tue May 12 19:44:34 2009 From: alandpearson at yahoo.com (Alan Pearson) Date: Tue, 12 May 2009 19:44:34 +0100 Subject: NSS Utils In-Reply-To: <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> References: <4A083EF7020000850005FD0B@FS-LIN-OES> <53899.81.149.45.143.1242128682.squirrel@www.massan.co.uk> <4A0919B3020000760000B0B4@gwia2.state.ut.us> <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> Message-ID: <5BADAE75-DC77-4B14-AFFB-86015B2E8330@yahoo.com> --- AlanP On 12 May 2009, at 15:08, Matt Smith wrote: > On 5/12/2009 at 9:25 AM, in message > <46084.88.211.54.85.1242134707.squirrel at 83.67.10.8>, "Alan Pearson" > wrote: >> This isn't really that helpful >> >> 1) Running on Linux it shows numeric userids, for users, not the > username >> 2) You can't get a report per user with a breakdown of what files are > in use >> >> Something that the users could access themselves would be the way > forward, >> where they could see what files they own, and the size each takes. >> >> Nearest I've seen is a tool called Disk Space Usage >> http://www.simonsware.com/downloads.html >> >> but it's not ideal and is slooow > > You know, you're right. It really isn't as useful on OES Linux. I > wonder if Novell plans on "fixing" that? > > Perhaps there's something in the mighty JRB utilities that would > suffice? I'd look but need to concentrate elsewhere right now. I am > interested in seeing where this threat leads though. > > -Matt > > -- > > Matt Smith Network Technology Specialist > Oconee County School System, Oconee County, Georgia > Office of Instruction and Technology 706-769-5685 x1314 > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From novell at jrbsoftware.com Tue May 12 23:27:23 2009 From: novell at jrbsoftware.com (John Baird) Date: Wed, 13 May 2009 10:27:23 +1200 Subject: NSS Utils In-Reply-To: <4A094A7F.E4C1.0068.0@oconee.k12.ga.us> References: <46084.88.211.54.85.1242134707.squirrel@83.67.10.8> Message-ID: <4A0AA08B.5683.AAF514@localhost> > You know, you're right. It really isn't as useful on OES Linux. I > wonder if Novell plans on "fixing" that? > > Perhaps there's something in the mighty JRB utilities that would > suffice? I'd look but need to concentrate elsewhere right now. I am > interested in seeing where this threat leads though. Whodidit in JRButils can scan a directory structure or an entire volume and show ownership and size for every file, or for only those entries owned by a particular user. It works on both NSS and NCP volumes on OES Linux but must be run from a Windows client. Other programs e.g. dquota, can show the quota and usage for a single user. John From joea at j4computers.com Tue May 12 23:47:48 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 12 May 2009 18:47:48 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A098115020000850005FD39@FS-LIN-OES> References: <4A098115020000850005FD39@FS-LIN-OES> Message-ID: <4A09C454020000850005FD3D@FS-LIN-OES> Upon reflection, Inherit_Postix_Permissions is really unnecessary, IMHO. Just that TIDS insist it should be invoked. Of greater concern is that I must be misreading the docs, or it really does not work in OES1, as the resource goes comatose when I attempt to bring it online. Is the key to use the "bind" command to bind a mounted volume to the NCP object? joe a. >>> On 5/12/2009 at 2:00 PM, "joea at j4computers.com" wrote: > What about "Inherit_POSIX_Permissions"? > > Will I be able to specify that using the ncpcon commands in OES1? > > Need to know this, before going much further, as it will be a show stopper. > > joe a. > >>>> "Tim Heywood" 05/12/09 12:55 PM >>> > Can't use the script - but otherwise it still works (or did on the 4 > clusters where we have sorted this before) > > HTH > > T > From joea at j4computers.com Wed May 13 16:29:05 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Wed, 13 May 2009 11:29:05 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0AAEF0020000850005FD46@FS-LIN-OES> Any feedback/input? Bind fails. Have tried with and without "--ncpservername=" and "--ipaddress=" prefix, with simple server name and contextual (NDS) server name. Tried creating ncpserver object and ncs.ncpserver objects and referencing each (one at a time) in bind. Does it simply not work? Someone have a working example and a description of how they created the ncpserver object? This is OES1. joe a. > >>>> "Tim Heywood" 05/12/09 12:55 PM >>> > Can't use the script - but otherwise it still works (or did on the 4 > clusters where we have sorted this before) > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Wed May 13 21:01:15 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Wed, 13 May 2009 16:01:15 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0AEEC6020000850005FD89@FS-LIN-OES> OK, so this is "working", now. But now that I am getting into it, it appears to be of limited utility. The bind appears to be "per server" and not "per resource". Not quite the solution he expected. So, if I have multiple resources mounted on a given server, I see all those resources, along with SYS and _ADMIN of that server. If a given resouce migrates to another node, it becomes "invisible", via the virtual server to which its former host was bound and can only be seen via its new host node, unless that new node is bound to yet another virtual server and one happens to look at that VS. So, where is the benefit I anticipated? That being, the ability to "map" or look at, a single point of reference, no matter where the resources actually reside. Unless I am miss focused here, it appears this is "useful" only for simple clusters. My admittedly limited exposure seems to indicate that one cannot execute multiple binds on a single server. Am I missing something? On this topic? Or am I guilty of "hearing what I wanted to hear" rather than what was being said? joe a. >>> "joea at j4computers.com" 05/13/09 11:32 AM >>> Any feedback/input? Bind fails. Have tried with and without "--ncpservername=" and "--ipaddress=" prefix, with simple server name and contextual (NDS) server name. Tried creating ncpserver object and ncs.ncpserver objects and referencing each (one at a time) in bind. Does it simply not work? Someone have a working example and a description of how they created the ncpserver object? This is OES1. joe a. > >>>> "Tim Heywood" 05/12/09 12:55 PM >>> > Can't use the script - but otherwise it still works (or did on the 4 > clusters where we have sorted this before) > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Wed May 13 21:13:45 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Wed, 13 May 2009 16:13:45 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0AEEC6020000850005FD89@FS-LIN-OES> References: <4A0AEEC6020000850005FD89@FS-LIN-OES> Message-ID: <4A0AF1B9020000750004D7E9@inet.eastcobbgroup.com> I haven't been following this thread closely, but it looks like you are trying to cluster a couple of OES1 servers? It's kind of a PITA do this with OES1, but it does work. The docs actually do tell you how to do it, if you work at putting them all together, along with some tids. We have had a few in production, way back when OES1 was all that was available. Like I said, I haven't been following too closely, because I haven't dealt with an OES1 server in a couple of years, and that was to migrate it to OES2. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/13/2009 04:01 PM >>> OK, so this is "working", now. But now that I am getting into it, it appears to be of limited utility. The bind appears to be "per server" and not "per resource". Not quite the solution he expected. So, if I have multiple resources mounted on a given server, I see all those resources, along with SYS and _ADMIN of that server. If a given resouce migrates to another node, it becomes "invisible", via the virtual server to which its former host was bound and can only be seen via its new host node, unless that new node is bound to yet another virtual server and one happens to look at that VS. So, where is the benefit I anticipated? That being, the ability to "map" or look at, a single point of reference, no matter where the resources actually reside. Unless I am miss focused here, it appears this is "useful" only for simple clusters. My admittedly limited exposure seems to indicate that one cannot execute multiple binds on a single server. Am I missing something? On this topic? Or am I guilty of "hearing what I wanted to hear" rather than what was being said? joe a. >>> "joea at j4computers.com" 05/13/09 11:32 AM >>> Any feedback/input? Bind fails. Have tried with and without "--ncpservername=" and "--ipaddress=" prefix, with simple server name and contextual (NDS) server name. Tried creating ncpserver object and ncs.ncpserver objects and referencing each (one at a time) in bind. Does it simply not work? Someone have a working example and a description of how they created the ncpserver object? This is OES1. joe a. > >>>> "Tim Heywood" 05/12/09 12:55 PM >>> > Can't use the script - but otherwise it still works (or did on the 4 > clusters where we have sorted this before) > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From tim at nds8.co.uk Wed May 13 21:16:03 2009 From: tim at nds8.co.uk (Tim Heywood) Date: Wed, 13 May 2009 21:16:03 +0100 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0AEEC6020000850005FD89@FS-LIN-OES> References: <4A0AEEC6020000850005FD89@FS-LIN-OES> Message-ID: <4A0B3893020000BB0005BE4D@mail2.nds8.com> LOL - In answer, I think the problem is "yes" you are doing something wrong. In principle, a Clustered resource is a single IP number that identifies that resource - moving that IP number (a secondary IP number) from one node to the other is the basis of a cluster. The load and unload scripts take the simple IP number, the mounting of the disk and then the binding of NCP into a simple script. Where I read that you have gone wrong is that you are looking at this as running on the node and hence the node's IP number - this is not the case. Each resource has it's own IP number and only that specified number should ever appear in any load or unload script for the resource. HTH T -- Tim Heywood NDS8 Novell Platinum Solution Provider Office: +44 (0) 131 538 8202 Mobile: +44 (0) 7974 134264 >>> On 13 May, 2009 at 21:01, "joea at j4computers.com" wrote: > OK, so this is "working", now. But now that I am getting into it, it appears > to be of limited utility. > > The bind appears to be "per server" and not "per resource". Not quite the > solution he expected. > > So, if I have multiple resources mounted on a given server, I see all those > resources, along with SYS and _ADMIN of that server. If a given resouce > migrates to another node, it becomes "invisible", via the virtual server to > which its former host was bound and can only be seen via its new host node, > unless that new node is bound to yet another virtual server and one happens > to look at that VS. > > So, where is the benefit I anticipated? That being, the ability to "map" or > look at, a single point of reference, no matter where the resources actually > reside. Unless I am miss focused here, it appears this is "useful" only for > simple clusters. > > My admittedly limited exposure seems to indicate that one cannot execute > multiple binds on a single server. > > Am I missing something? On this topic? Or am I guilty of "hearing what I > wanted to hear" rather than what was being said? > > joe a. > > >>>> "joea at j4computers.com" 05/13/09 11:32 AM >>> > Any feedback/input? > > Bind fails. Have tried with and without "--ncpservername=" and "--ipaddress=" > prefix, with simple server name and contextual (NDS) server name. Tried > creating ncpserver object and ncs.ncpserver objects and referencing each (one > at a time) in bind. > > Does it simply not work? Someone have a working example and a description of > how they created the ncpserver object? > > This is OES1. > > joe a. > >> >>>>> "Tim Heywood" 05/12/09 12:55 PM >>> >> Can't use the script - but otherwise it still works (or did on the 4 >> clusters where we have sorted this before) >> >> HTH >> >> T >> > > > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > ***Scanned by M+ Guardian*** The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. From joea at j4computers.com Thu May 14 14:29:47 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 09:29:47 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0BE48C020000850005FDA3@FS-LIN-OES> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > From joea at j4computers.com Thu May 14 15:51:58 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 10:51:58 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0BF7C5020000850005FDB0@FS-LIN-OES> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Thu May 14 16:09:59 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 14 May 2009 11:09:59 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0BF7C5020000850005FDB0@FS-LIN-OES> References: <4A0BF7C5020000850005FDB0@FS-LIN-OES> Message-ID: <4A0BFC070200007500032A59@inet.eastcobbgroup.com> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Thu May 14 16:30:13 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 11:30:13 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0C00C5020000850005FDC0@FS-LIN-OES> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Thu May 14 17:07:21 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 14 May 2009 12:07:21 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0C00C5020000850005FDC0@FS-LIN-OES> References: <4A0C00C5020000850005FDC0@FS-LIN-OES> Message-ID: <4A0C09790200007500032A71@inet.eastcobbgroup.com> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Thu May 14 17:59:53 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 12:59:53 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0C15CA020000850005FDE2@FS-LIN-OES> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Thu May 14 19:19:32 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 14 May 2009 14:19:32 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0C15CA020000850005FDE2@FS-LIN-OES> References: <4A0C15CA020000850005FDE2@FS-LIN-OES> Message-ID: <4A0C28740200007500031D33@inet.eastcobbgroup.com> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Thu May 14 19:49:54 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 14:49:54 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0C2F7E020000850005FE00@FS-LIN-OES> Ah. I am still not being clear. I am not asking for the resources to "exist" on two nodes at the same time. Clearly, they can "exist" on only one node at a time. What I am asking for is a means to "map" the resource(s) to a single object, an "ncpserver" object ("My_NCP", say) and have that object "see" the resource as being attched to it (ncpserver object) regardless of which node the resource exists on, at the moment. The intent/desire being, a single point of reference, to the resource/volume, that being: "My_NCP". So, if a resource/volume migrates to another node, one does not have to know where the resouce moved to, but simply still references that single object, "My_NCP" and "sees" the resource/volume. I was under the impression, from various sources, that this was easily done in NetWare NCS and could be done, in Linux NCS, with a bit of effort. That is, a "Virtual Server" existed, that represented "The Cluster" and that, regardless of what node the resouces existed on at the moment, they always appeared under "The Cluster" (or "Virtual Server"). This is not so, apparently. joe a. >>> "James Taylor" 05/14/09 2:20 PM >>> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Thu May 14 20:08:47 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 14 May 2009 15:08:47 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0C2F7E020000850005FE00@FS-LIN-OES> References: <4A0C2F7E020000850005FE00@FS-LIN-OES> Message-ID: <4A0C33FF0200007500031D51@inet.eastcobbgroup.com> You don't "manually" create an ncp object and associate a resource with it. You create a cluster resource, in this case a cluster virtual server, which has the resource as a component of it. Once you do that, the object can be moved to any node in the cluster. You cannot create an ncp object and then make it a cluster virtual server, it has to be created that way to begin with. -jt >>> "joea at j4computers.com" 5/14/2009 02:49 PM >>> Ah. I am still not being clear. I am not asking for the resources to "exist" on two nodes at the same time. Clearly, they can "exist" on only one node at a time. What I am asking for is a means to "map" the resource(s) to a single object, an "ncpserver" object ("My_NCP", say) and have that object "see" the resource as being attched to it (ncpserver object) regardless of which node the resource exists on, at the moment. The intent/desire being, a single point of reference, to the resource/volume, that being: "My_NCP". So, if a resource/volume migrates to another node, one does not have to know where the resouce moved to, but simply still references that single object, "My_NCP" and "sees" the resource/volume. I was under the impression, from various sources, that this was easily done in NetWare NCS and could be done, in Linux NCS, with a bit of effort. That is, a "Virtual Server" existed, that represented "The Cluster" and that, regardless of what node the resouces existed on at the moment, they always appeared under "The Cluster" (or "Virtual Server"). This is not so, apparently. joe a. >>> "James Taylor" 05/14/09 2:20 PM >>> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From seuferer at alan.ameslab.gov Thu May 14 20:12:45 2009 From: seuferer at alan.ameslab.gov (David Seuferer) Date: Thu, 14 May 2009 14:12:45 -0500 Subject: Migrating rights from eDir to eDir Message-ID: <4A0C26DD020000830000654A@alan.ameslab.gov> Hello, We are in the process of migrating from our old tree, Netware based, to a new tree, Suse based, and I could have sworn there was a tool that would mirgate eDir rights from one tree to another, but I can't find it anywhere. Any ideas on where I can find this? Thanks! David Seuferer, CNE6 Ames Laboratory Information Systems seuferer at ameslab.gov fax: (515) 294-5638 phone: (515) 294-6053 "If you can dream it, you can do it. Always remember that this whole thing was started with a dream and a mouse." - Walt Disney From joea at j4computers.com Thu May 14 20:25:48 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Thu, 14 May 2009 15:25:48 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0C37FA020000850005FE13@FS-LIN-OES> Then perhaps you would be so kind as to point me to the exact documents that describe how to accomplish this? The OES docs I followed seemed quite clear that one created (Manually) an NCP server object,using iManager. Docs then describes this as a "virtual" server, to which one "binds". Thanks for your help. joe a. >>>> Taylor" 05/14/09 3:10 PM >>> You don't "manually" create an ncp object and associate a resource with it. You create a cluster resource, in this case a cluster virtual server, which has the resource as a component of it. Once you do that, the object can be moved to any node in the cluster. You cannot create an ncp object and then make it a cluster virtual server, it has to be created that way to begin with. -jt >>> "joea at j4computers.com" 5/14/2009 02:49 PM >>> Ah. I am still not being clear. I am not asking for the resources to "exist" on two nodes at the same time. Clearly, they can "exist" on only one node at a time. What I am asking for is a means to "map" the resource(s) to a single object, an "ncpserver" object ("My_NCP", say) and have that object "see" the resource as being attched to it (ncpserver object) regardless of which node the resource exists on, at the moment. The intent/desire being, a single point of reference, to the resource/volume, that being: "My_NCP". So, if a resource/volume migrates to another node, one does not have to know where the resouce moved to, but simply still references that single object, "My_NCP" and "sees" the resource/volume. I was under the impression, from various sources, that this was easily done in NetWare NCS and could be done, in Linux NCS, with a bit of effort. That is, a "Virtual Server" existed, that represented "The Cluster" and that, regardless of what node the resouces existed on at the moment, they always appeared under "The Cluster" (or "Virtual Server"). This is not so, apparently. joe a. >>> "James Taylor" 05/14/09 2:20 PM >>> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joe.doupnik at oucs.ox.ac.uk Thu May 14 20:32:54 2009 From: joe.doupnik at oucs.ox.ac.uk (jrd) Date: Thu, 14 May 2009 20:32:54 +0100 Subject: Migrating rights from eDir to eDir In-Reply-To: <4A0C26DD020000830000654A@alan.ameslab.gov> References: <4A0C26DD020000830000654A@alan.ameslab.gov> Message-ID: <4A0C71E6.5060800@oucs.ox.ac.uk> David Seuferer wrote: > Hello, > We are in the process of migrating from our old tree, Netware > based, to a new tree, Suse based, and I could have sworn there was > a tool that would mirgate eDir rights from one tree to another, > but I can't find it anywhere. Any ideas on where I can find this? > > > > Thanks! > > > > > David Seuferer, CNE6 > Ames Laboratory > Information Systems ------------ For OES2 SP1/Linux, yes; for SLES itself, no. If your target is OES2 SP1/Linux then please read up on its migration facilities (GUI and many command line utils). Joe D. From James.Taylor at eastcobbgroup.com Thu May 14 20:33:26 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 14 May 2009 15:33:26 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0C37FA020000850005FE13@FS-LIN-OES> References: <4A0C37FA020000850005FE13@FS-LIN-OES> Message-ID: <4A0C39C60200007500031D6A@inet.eastcobbgroup.com> OK, so there is a manual configuration of the ncp object if the resource is a linux native (posix) partition. I'll defer to others on that who may hove more recent experience dealing with it. -jt >>> "joea at j4computers.com" 5/14/2009 03:25 PM >>> Then perhaps you would be so kind as to point me to the exact documents that describe how to accomplish this? The OES docs I followed seemed quite clear that one created (Manually) an NCP server object,using iManager. Docs then describes this as a "virtual" server, to which one "binds". Thanks for your help. joe a. >>>> Taylor" 05/14/09 3:10 PM >>> You don't "manually" create an ncp object and associate a resource with it. You create a cluster resource, in this case a cluster virtual server, which has the resource as a component of it. Once you do that, the object can be moved to any node in the cluster. You cannot create an ncp object and then make it a cluster virtual server, it has to be created that way to begin with. -jt >>> "joea at j4computers.com" 5/14/2009 02:49 PM >>> Ah. I am still not being clear. I am not asking for the resources to "exist" on two nodes at the same time. Clearly, they can "exist" on only one node at a time. What I am asking for is a means to "map" the resource(s) to a single object, an "ncpserver" object ("My_NCP", say) and have that object "see" the resource as being attched to it (ncpserver object) regardless of which node the resource exists on, at the moment. The intent/desire being, a single point of reference, to the resource/volume, that being: "My_NCP". So, if a resource/volume migrates to another node, one does not have to know where the resouce moved to, but simply still references that single object, "My_NCP" and "sees" the resource/volume. I was under the impression, from various sources, that this was easily done in NetWare NCS and could be done, in Linux NCS, with a bit of effort. That is, a "Virtual Server" existed, that represented "The Cluster" and that, regardless of what node the resouces existed on at the moment, they always appeared under "The Cluster" (or "Virtual Server"). This is not so, apparently. joe a. >>> "James Taylor" 05/14/09 2:20 PM >>> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From Tim at nds8.co.uk Thu May 14 21:00:09 2009 From: Tim at nds8.co.uk (Tim Heywood) Date: Thu, 14 May 2009 21:00:09 +0100 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0C8659020000BB00014703@mail2.nds8.com> References: <4A0C8659020000BB00014700@mail2.nds8.com> <4A0C8659020000BB00014703@mail2.nds8.com> Message-ID: <4A0C8659020000BB00014703@mail2.nds8.com> Joe, I think that I'm with James here. You really need to go back and look at this from first principles. A node is a node and other than that has no baring on the script. Let us say that the nodes have an ip of 192.168.0.10. 11 and 12 The resource and the vncp server should have a common ip (say 0.66) The load and unload scripts should only ever reference 66. The mapped drive should only reference the vncp server (though during testing you can use the 66 ip) Because the 66 is a secondary address, if it moves from one node to another the clients follow The fact that you can see the SYS and _admin is a by-product of NSS and NCP and should never be referenced from the vncp server That help? T -----Original Message----- From: "joea at j4computers.com" To: Sent: 14/05/2009 19:49:54 Subject: Re: Map to cluster volume on Linux (ncp) Ah. I am still not being clear. I am not asking for the resources to "exist" on two nodes at the same time. Clearly, they can "exist" on only one node at a time. What I am asking for is a means to "map" the resource(s) to a single object, an "ncpserver" object ("My_NCP", say) and have that object "see" the resource as being attched to it (ncpserver object) regardless of which node the resource exists on, at the moment. The intent/desire being, a single point of reference, to the resource/volume, that being: "My_NCP". So, if a resource/volume migrates to another node, one does not have to know where the resouce moved to, but simply still references that single object, "My_NCP" and "sees" the resource/volume. I was under the impression, from various sources, that this was easily done in NetWare NCS and could be done, in Linux NCS, with a bit of effort. That is, a "Virtual Server" existed, that represented "The Cluster" and that, regardless of what node the resouces existed on at the moment, they always appeared under "The Cluster" (or "Virtual Server"). This is not so, apparently. joe a. >>> "James Taylor" 05/14/09 2:20 PM >>> To the best of my understanding of what you are asking, it doesn't look like what you are trying to do is an NCS environment. I can't see how what you are asking about relates to clustering. If the resources (volumes?) have been successfully proven to be able to move between nodes in an NCS environment, then they are already NCP server objects. If you are referring to being able to connect them to one node or another, then I assume they are partitions of some kind on shared storage, in which case, you would create cluster virtual server objects which include one or more of those resources, and which would be able to migrate between nodes, but would not exist on more than one node at a time. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 12:59 PM >>> I am sorry I have not been more clear. The objects I was speaking of are the NCP objects, which do have to be manually created, in this environment. I have at least a basic understanding of clustering, having built a few and having done some work with startup and shutdown scripts. And, when required, have applied channel delivered updates to nodes, which requires offline and online of resources, cluster leaves and reboots. FWIW, I no longer "migrate" resources due to some odd happenings when doing so. My exposure to this particular facet of NCS is limited. While I understand this may be getting tedious, what I am looking for is a yes or no answer to my supposition. As I understand what a variety of people have been saying, on and off list, this was readily accomplished under NetWare. Never had the pleasure, myself. So, let's presume a functional OES1 two node cluster. Virgins On Linux. The resources we are concerned with use non-NSS volumes. Configured with EVMS. Never been on NetWare. The resources have been proven to move between nodes, without problems. "The Cluster(s) work". Been working, as is, for over a year. Can one: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). (By "resouce" I mean the "thing" that one assigns startup and shutdown scripts, via iManager) Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). NCS can either do this, or it cannot. joe a. >>> "James Taylor" 05/14/09 12:08 PM >>> Joe - You don't manually create any of the cluster objects outside of the using the cluster plugins. The most basic thing that needs to be done, after the cluster object is created and the nodes are functioning within the cluster, is to create a cluster virtual server object, which needs an already existing NSS Pool object. This can also be done with a non-NSS NCP partition, but it's a bit more involved. To be honest, I haven't been following the details of what you have been posting because there doesn't seem to be any starting point to getting back to where you need to be with a cluster virtual server. You seem to be missing the point on how NCS works. Resources don't exist on nodes. They are independent of nodes, but can be assigned to *individual* nodes, but cannot exist on more than one node at any particular time. Resources associated with a cluster virtual server (which is classified as an NCP object) all have to be on the same node at the same time. You never have resources associated with one cluster virtual server on more than one node at a time. That's why I am suggesting you review the documentation and tids in great detail, then start from scratch. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 11:30 AM >>> James, Thanks for the response. I would certainly think this should be a basic need/function. To get back on track, I believe you are saying it *should* work the way I've described? However ineptly? One should be able to: Have a "resource" on two (or more) different cluster "nodes" ("host" servers). Have a single object (manually created "ncp server") Have each of those "resources" "mapped" ("bind"-ed) that single object (manually created "ncp server"). yes? joe a. >>> "James Taylor" 05/14/09 11:11 AM >>> Joe - You really seem to be so far off-track with how you are going about this that I don't think you have any chance of even finding the track again, much less getting back on it. I would highly suggest that you trash the set up you have and review all of the documentation carefully, then start from the beginning to make sure that you have the basic configuration down. What you are trying to do is the most basic functionality of clustering and, even on OES1, this is pretty straightforward. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/14/2009 10:51 AM >>> I have to amend part of my rantings. unbind does appear to unbind. I was looking at an odd object I had created early one an "ncs.ncp" object. That seems to be an odd creature. In any event, it still appears that, once a resource is bound to an object, cannot bind a resource from a different node, to that object. Nor and one move that resource to another node, "bind" it *from* there, and have it seen under the object. And to our thinking, that is the whole point, to be able to show a resouce, regardless of where it actually lives, as being under the same object, no matter what. Ideally, resources on different nodes, should be able to be seen under the same object. And move about, while still being seen there. With some allowance for the time needed to sort things out. That does not seem to be happening. joe a. >>> "joea at j4computers.com" 05/14/09 9:31 AM >>> Glad I could brighten your day. I may have used incorrect terminology, or not expressed myself properly. No, really. Don't think it matters, in this description, if OES1 or OES2. Concept is the same. My goal is to be able to reference a single object, regardless of which node the volumes (resource) reside upon. There are at least 3 volumes, resources, per cluster. For reasons best left for another thread, they are, usually, maintained on one node of the cluster. So, I want to be able to migrate these volumes (resources) to the other node and still see them on the single object ("virtual" ncp server). That does not seem possible, in my limited experience. I hope, beyond hope, to be wrong. When I say "ncpcon bind --ncpservername=my_name --ipaddress=IP_address_of_resource" then - SYS and _ADMIN of whatever node the resouce is on become visible. Apparently. At this point, there are no non stock volumes mounted. If I then say "ncpcon mount volume (or ncpcon create volume path)" then that volume becomes visible as well. If I then say similar sweet things about other volumes, they become visible under that virtiual server (my_name). I guess that's fine. However, if I then attempt to "unbind", the command apparenlty completes correctly, but it seems it never actually unbinds. The base volumes and any extra mounted volumes are still visible. I can unmount those and they disappear from "my_name". I can then delete them. A second unbind fails, as there is nothing bound. Supposedly. However, the SYS and _Admin volumes are still visible. I know the command for del_secondary_IP was issued, for all resources. I can then say rcndsd restart and, virtually immediately, the SYS and _ADMIN volumes disappear, if I re-enter "my_name". As soon as eDir is up again, and I refresh "my_name", yep, SYS and _ADMIN are back. I know it is the same node as I dropped a node name file is SYS, just for easy ID purposes. That does not seem like correct behavior. It may be extraneous to add that I can issue a "bind" on the other node, which appears to execute without error, but, of course, it seems to have no effect on "my_name". joe a. >>> On 5/13/2009 at 4:16 PM, "Tim Heywood" wrote: > LOL - In answer, I think the problem is "yes" you are doing something > wrong. > > In principle, a Clustered resource is a single IP number that > identifies that resource - moving that IP number (a secondary IP number) > from one node to the other is the basis of a cluster. > > The load and unload scripts take the simple IP number, the mounting of > the disk and then the binding of NCP into a simple script. Where I read > that you have gone wrong is that you are looking at this as running on > the node and hence the node's IP number - this is not the case. Each > resource has it's own IP number and only that specified number should > ever appear in any load or unload script for the resource. > > HTH > > T > _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell ***Scanned by M+ Guardian*** The information contained in this email is intended for the person to whom it is addressed and may contain confidential and/or privileged information. You should not copy, retain, forward or disclose its contents to anyone else, or take any action based upon it, if it is not addressed to you personally. If the message is received by anyone other than the addressee, please notify the sender and delete the message. NDS8 does not accept responsibility for changes made to this message after it was sent. Whilst all reasonable care has been taken to avoid the transmission of viruses, it is the responsibility of the recipient to ensure that the onward transmission, opening or use of this message and any attachments will not adversely affect its systems or data. From joea at j4computers.com Fri May 15 15:00:10 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 15 May 2009 10:00:10 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0D3D29020000850005FE59@FS-LIN-OES> This is an an attempt to backup and regroup. If we get nowhere with this go round, I will attempt to remain silent on the topic. If I understand things, and I believe I do, I have had this functional, since, almost, the beginning. When I bind the VNCP server to the "resource" IP, whatever volumes are mounted show up under the VNCP object. The problem appears to be that, once a resource IP is bound, I see all the mounted volumes. Not speaking of the SYS and _ADMIN volumes, but of any volumes other resources may have mounted on that node. For example, I have a two node cluster. Resources R1 and R2 are loaded on node "N1". Each resource has a volume created and mounted at load time, V1 and V2. When I bind either resource IP to the VNCP name, both volumes, V1 and V2, appear. Now, if I dismount V1, V1 disappears from the view of VNCP. If I mount V1, V1 reappears in VNCP. If I migrate R1 to N2, V1 also disappears from VNCP. As one might expect. However, if I then attempt to bind R2 to VNCP, the bind completes without error report, but, V1 never reappears in VNCP. Is that normal behaviour? If so, that seems a serious limitation, for those who have more than one resource on a cluster. If not, goodness, I am doing something wrong. Or, Horrors, there is a bug. joe a. From joea at j4computers.com Fri May 15 20:21:05 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 15 May 2009 15:21:05 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A0D885C020000850005FE98@FS-LIN-OES> One last, (I swear) observation. I create two VNCP objects, VNCP1 and VNCP2. Load Resources R1 and R1 on node N1. Have bind statements like this: (R1 ) - "ncpcon bind --ncpservername=VNCP1 --ipaddress=R1_IP_address" (R2 ) - "ncpcon bind --ncpservername=VNCP2 --ipaddress=R2_IP_address" The volumes associated with each resource show up under VNCP1, VNCP2 is empty. Migrate R1 to node N2. The volume associated with R1 moves to VNCP2. Migrate R2 to node N2. The volume associated with R2 moves to VNCP2, *and* VNCP1 is blank. If I migrate both resources back to node N1, VNCP2 goes empty and the volumes show up in VNCP1. That is not the behavior I have been told existed on NetWare. Is that how it works in OES2? In any event it does not seem like correct behavior. It appears that only a single bind can exert influence on any given node, any other binds are either ignored, or bungle things. So, the idea that there is one VNCP object per "resource" does not seem to hold up. End Of Transmission. joe a. >>> "joea at j4computers.com" 05/15/09 10:02 AM >>> This is an an attempt to backup and regroup. If we get nowhere with this go round, I will attempt to remain silent on the topic. If I understand things, and I believe I do, I have had this functional, since, almost, the beginning. When I bind the VNCP server to the "resource" IP, whatever volumes are mounted show up under the VNCP object. The problem appears to be that, once a resource IP is bound, I see all the mounted volumes. Not speaking of the SYS and _ADMIN volumes, but of any volumes other resources may have mounted on that node. For example, I have a two node cluster. Resources R1 and R2 are loaded on node "N1". Each resource has a volume created and mounted at load time, V1 and V2. When I bind either resource IP to the VNCP name, both volumes, V1 and V2, appear. Now, if I dismount V1, V1 disappears from the view of VNCP. If I mount V1, V1 reappears in VNCP. If I migrate R1 to N2, V1 also disappears from VNCP. As one might expect. However, if I then attempt to bind R2 to VNCP, the bind completes without error report, but, V1 never reappears in VNCP. Is that normal behaviour? If so, that seems a serious limitation, for those who have more than one resource on a cluster. If not, goodness, I am doing something wrong. Or, Horrors, there is a bug. joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From bbrush at gmail.com Fri May 15 21:41:32 2009 From: bbrush at gmail.com (Bill Brush) Date: Fri, 15 May 2009 15:41:32 -0500 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0D885C020000850005FE98@FS-LIN-OES> References: <4A0D885C020000850005FE98@FS-LIN-OES> Message-ID: <167f4090905151341h4972ccfra8cba2bfaeee4ede@mail.gmail.com> Ok, I'll be honest, I don't know that I've completely followed your explanation, but I will explain what I see on Netware. For convenience sake let's say I have 3 nodes Alpha, Beta, Delta. I have 3 clustered volumes, Home, Shared, and Public. There are 3 virtual server names to go with them, Home-server, Shared-server, and public-server. If I arrange things so that we have one virtual server residing on each physical node, then someone browsing to map a drive will only see one volume on each node. So let's say that Home-server is the resource that is loaded on node Alpha. Someone browsing to Home-server for creating a mapped drive will see "Home" as the only volume. Likewise if someone browses to "Alpha" to map a drive then they will see "Home." The server can't partition the volumes it is serving by address, it shows all of them regardless of which address someone requests information from. Now we migrate Shared-server to node Alpha. Node alpha now has "Home", and "Shared" for volumes, and three ip addresses (Alpha, Shared-server, and Home-server's addresses.) Someone browsing to "Home-server" to map a drive will see Home and Shared as volumes which can be browsed. Likewise someone browsing "Shared-server" or "Alpha" will see all the volumes. In the event of a migration of a resource, it dismounts the volume, deactivates the pool, removes the secondary IP address, then the target physical node activates the pool, mounts the volume and adds the IP address. If someone accesses "Home" via the "Home-server" address, it won't matter which physical node it's mounted on, it will maintain the connection. If someone was to access "Shared" via the "Home-server" address, while they're both on the same physical node it would work, UNTIL "Home-server" is moved to another physical node. >From what your describing it "almost" works, but I'm not familiar enough with NCS on Linux to say where it isn't doing what it's supposed to. Generally speaking you'll have one virtual server per cluster resource. I've played around with having more than one volume per virtual server, but it didn't really like it. Bill On Fri, May 15, 2009 at 2:21 PM, joea at j4computers.com wrote: > One last, (I swear) observation. > > I create two VNCP objects, VNCP1 and VNCP2. > > Load Resources R1 and R1 on node N1. Have bind statements like this: > > (R1 ) - "ncpcon bind --ncpservername=VNCP1 --ipaddress=R1_IP_address" > (R2 ) - "ncpcon bind --ncpservername=VNCP2 --ipaddress=R2_IP_address" > > The volumes associated with each resource show up under VNCP1, VNCP2 is > empty. > Migrate R1 to node N2. The volume associated with R1 moves to VNCP2. > Migrate R2 to node N2. The volume associated with R2 moves to VNCP2, *and* > VNCP1 is blank. > > If I migrate both resources back to node N1, VNCP2 goes empty and the > volumes show up > in VNCP1. > > That is not the behavior I have been told existed on NetWare. Is that how > it works in OES2? > > In any event it does not seem like correct behavior. It appears that only a > single bind can > exert influence on any given node, any other binds are either ignored, or > bungle things. > > So, the idea that there is one VNCP object per "resource" does not seem to > hold up. > > End Of Transmission. > > joe a. > > >>> "joea at j4computers.com" 05/15/09 10:02 AM >>> > This is an an attempt to backup and regroup. If we get nowhere with this > go round, I will attempt to remain silent on the topic. > > If I understand things, and I believe I do, I have had this functional, > since, > almost, the beginning. > > When I bind the VNCP server to the "resource" IP, whatever volumes are > mounted show up under the VNCP object. > > The problem appears to be that, once a resource IP is bound, I see all the > mounted volumes. Not speaking of the SYS and _ADMIN volumes, but of > any volumes other resources may have mounted on that node. > > For example, I have a two node cluster. Resources R1 and R2 are loaded > on node "N1". Each resource has a volume created and mounted at load > time, V1 and V2. When I bind either resource IP to the VNCP name, both > volumes, V1 and V2, appear. > > Now, if I dismount V1, V1 disappears from the view of VNCP. If I mount V1, > V1 reappears in VNCP. > > If I migrate R1 to N2, V1 also disappears from VNCP. As one might expect. > > However, if I then attempt to bind R2 to VNCP, the bind completes without > error report, but, V1 never reappears in VNCP. > > Is that normal behaviour? > > If so, that seems a serious limitation, for those who have more than one > resource on a cluster. > > If not, goodness, I am doing something wrong. Or, Horrors, there is a bug. > > joe a. > > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > From James.Taylor at eastcobbgroup.com Fri May 15 22:23:53 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Fri, 15 May 2009 17:23:53 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A0D885C020000850005FE98@FS-LIN-OES> References: <4A0D885C020000850005FE98@FS-LIN-OES> Message-ID: <4A0DA52A0200007500032DCD@inet.eastcobbgroup.com> If I am understanding what you are saying, you are trying to migrate resources independent of the cluster virtual servers that they are associated with. This would be invalid. resources have unique virtual cluster server objects that they are associated with, and the cluster virtual server objects are migrated to/from or onlined/offlined on a cluster node (host) as a unit. When you migrate R1 to N2, then VNCP1 will be on N1. R1 should never have any direct connection with VNCP2. If R1 and R2 happen to be on the same node, then you will be able to see all of the volumes associated with each resource from each virtual cluster server and the the node they are on, but you should only access the resources via the virtual cluster server IP address. If you do it any other way, then connections will fail when you migrate the resources. Cluster resources should never be accessed (except for specific administration purposes) by any other method than the virtual cluster server address. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/15/2009 03:21 PM >>> One last, (I swear) observation. I create two VNCP objects, VNCP1 and VNCP2. Load Resources R1 and R1 on node N1. Have bind statements like this: (R1 ) - "ncpcon bind --ncpservername=VNCP1 --ipaddress=R1_IP_address" (R2 ) - "ncpcon bind --ncpservername=VNCP2 --ipaddress=R2_IP_address" The volumes associated with each resource show up under VNCP1, VNCP2 is empty. Migrate R1 to node N2. The volume associated with R1 moves to VNCP2. Migrate R2 to node N2. The volume associated with R2 moves to VNCP2, *and* VNCP1 is blank. If I migrate both resources back to node N1, VNCP2 goes empty and the volumes show up in VNCP1. That is not the behavior I have been told existed on NetWare. Is that how it works in OES2? In any event it does not seem like correct behavior. It appears that only a single bind can exert influence on any given node, any other binds are either ignored, or bungle things. So, the idea that there is one VNCP object per "resource" does not seem to hold up. End Of Transmission. joe a. >>> "joea at j4computers.com" 05/15/09 10:02 AM >>> This is an an attempt to backup and regroup. If we get nowhere with this go round, I will attempt to remain silent on the topic. If I understand things, and I believe I do, I have had this functional, since, almost, the beginning. When I bind the VNCP server to the "resource" IP, whatever volumes are mounted show up under the VNCP object. The problem appears to be that, once a resource IP is bound, I see all the mounted volumes. Not speaking of the SYS and _ADMIN volumes, but of any volumes other resources may have mounted on that node. For example, I have a two node cluster. Resources R1 and R2 are loaded on node "N1". Each resource has a volume created and mounted at load time, V1 and V2. When I bind either resource IP to the VNCP name, both volumes, V1 and V2, appear. Now, if I dismount V1, V1 disappears from the view of VNCP. If I mount V1, V1 reappears in VNCP. If I migrate R1 to N2, V1 also disappears from VNCP. As one might expect. However, if I then attempt to bind R2 to VNCP, the bind completes without error report, but, V1 never reappears in VNCP. Is that normal behaviour? If so, that seems a serious limitation, for those who have more than one resource on a cluster. If not, goodness, I am doing something wrong. Or, Horrors, there is a bug. joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Mon May 18 14:41:10 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Mon, 18 May 2009 09:41:10 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A112D31020000850005FEB7@FS-LIN-OES> Gentlemen, all, I thank you for your efforts. It appears, after all, that I have it working "as designed" but not "as expected" (by me). To my simple mind, the major reason for having a "virtual server" is to allow resources to be migrated, node to node, without having to be concerned about where it is currently living. A user with a shortcut or mapped drive, or url pointer, to "R1" will not have to care if that resource gets migrated to another node. To my simple mind, the current functionality does not provide "high reliability" "connectivity". It simply provides an "alias" to whatever is mounted on it's associated node. I suppose an "enhancement request" is in order. joe a. >>> "James Taylor" 05/15/09 5:25 PM >>> If I am understanding what you are saying, you are trying to migrate resources independent of the cluster virtual servers that they are associated with. This would be invalid. resources have unique virtual cluster server objects that they are associated with, and the cluster virtual server objects are migrated to/from or onlined/offlined on a cluster node (host) as a unit. When you migrate R1 to N2, then VNCP1 will be on N1. R1 should never have any direct connection with VNCP2. If R1 and R2 happen to be on the same node, then you will be able to see all of the volumes associated with each resource from each virtual cluster server and the the node they are on, but you should only access the resources via the virtual cluster server IP address. If you do it any other way, then connections will fail when you migrate the resources. Cluster resources should never be accessed (except for specific administration purposes) by any other method than the virtual cluster server address. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/15/2009 03:21 PM >>> One last, (I swear) observation. I create two VNCP objects, VNCP1 and VNCP2. Load Resources R1 and R1 on node N1. Have bind statements like this: (R1 ) - "ncpcon bind --ncpservername=VNCP1 --ipaddress=R1_IP_address" (R2 ) - "ncpcon bind --ncpservername=VNCP2 --ipaddress=R2_IP_address" The volumes associated with each resource show up under VNCP1, VNCP2 is empty. Migrate R1 to node N2. The volume associated with R1 moves to VNCP2. Migrate R2 to node N2. The volume associated with R2 moves to VNCP2, *and* VNCP1 is blank. If I migrate both resources back to node N1, VNCP2 goes empty and the volumes show up in VNCP1. That is not the behavior I have been told existed on NetWare. Is that how it works in OES2? In any event it does not seem like correct behavior. It appears that only a single bind can exert influence on any given node, any other binds are either ignored, or bungle things. So, the idea that there is one VNCP object per "resource" does not seem to hold up. End Of Transmission. joe a. >>> "joea at j4computers.com" 05/15/09 10:02 AM >>> This is an an attempt to backup and regroup. If we get nowhere with this go round, I will attempt to remain silent on the topic. If I understand things, and I believe I do, I have had this functional, since, almost, the beginning. When I bind the VNCP server to the "resource" IP, whatever volumes are mounted show up under the VNCP object. The problem appears to be that, once a resource IP is bound, I see all the mounted volumes. Not speaking of the SYS and _ADMIN volumes, but of any volumes other resources may have mounted on that node. For example, I have a two node cluster. Resources R1 and R2 are loaded on node "N1". Each resource has a volume created and mounted at load time, V1 and V2. When I bind either resource IP to the VNCP name, both volumes, V1 and V2, appear. Now, if I dismount V1, V1 disappears from the view of VNCP. If I mount V1, V1 reappears in VNCP. If I migrate R1 to N2, V1 also disappears from VNCP. As one might expect. However, if I then attempt to bind R2 to VNCP, the bind completes without error report, but, V1 never reappears in VNCP. Is that normal behaviour? If so, that seems a serious limitation, for those who have more than one resource on a cluster. If not, goodness, I am doing something wrong. Or, Horrors, there is a bug. joe a. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Mon May 18 14:58:48 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Mon, 18 May 2009 09:58:48 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A112D31020000850005FEB7@FS-LIN-OES> References: <4A112D31020000850005FEB7@FS-LIN-OES> Message-ID: <4A1131580200007500032FEF@inet.eastcobbgroup.com> James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "joea at j4computers.com" 5/18/2009 09:41 AM >>> Gentlemen, all, I thank you for your efforts. ================================================================================================ To my simple mind, the major reason for having a "virtual server" is to allow resources to be migrated, node to node, without having to be concerned about where it is currently living. A user with a shortcut or mapped drive, or url pointer, to "R1" will not have to care if that resource gets migrated to another node. ================================================================================================= This is exactly the way it works. A user mapped to a virtual resource does not have to care. All mappings should be done based on virtual cluster server eDirectory objects. The local resources will not be visible as virtual cluster server eDir objects. The only way those could be incorrectly mapped is if you used server mapped volumes as opposed to eDirectory mapped volumes. Cluster objects only exists as eDir objects. None of the local node resources will show up as eDir cluster resources. ================================================================================================= To my simple mind, the current functionality does not provide "high reliability" "connectivity". It simply provides an "alias" to whatever is mounted on it's associated node. ================================================================================================= That is incorrect. The fact that virtual server resources are visible at the physical node level, and physical resources are visible at the virtual server level has nothing to do with the functionality. It is just a function of the way NCP servers work. If the cluster is configured and managed properly, using eDir objects for mapping this is not an issue for the users at all. ================================== I suppose an "enhancement request" is in order. ================================== Not really. -jt From joea at j4computers.com Tue May 19 03:00:51 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Mon, 18 May 2009 22:00:51 -0400 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A1131580200007500032FEF@inet.eastcobbgroup.com> References: <4A112D31020000850005FEB7@FS-LIN-OES> <4A1131580200007500032FEF@inet.eastcobbgroup.com> Message-ID: <4A11DA93020000850005FEDB@FS-LIN-OES> > > This is exactly the way it works. A user mapped to a virtual resource does > not have to care. All mappings should be done based on virtual cluster > server eDirectory objects. The local resources will not be visible as > virtual cluster server eDir objects. The only way those could be incorrectly > mapped is if you used server mapped volumes as opposed to eDirectory mapped > volumes. Cluster objects only exists as eDir objects. None of the local > node resources will show up as eDir cluster resources. > That may be the way you see it working on NetWare, but it does not seem to work that way in Linux OES1. It works the way I described it, at least on our gear. And the way I interpret Bill's response, it differs from what you describe. joe a. From petervl at gmail.com Tue May 19 03:13:20 2009 From: petervl at gmail.com (Peter Van Lone) Date: Mon, 18 May 2009 21:13:20 -0500 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A11DA93020000850005FEDB@FS-LIN-OES> References: <4A112D31020000850005FEB7@FS-LIN-OES> <4A1131580200007500032FEF@inet.eastcobbgroup.com> <4A11DA93020000850005FEDB@FS-LIN-OES> Message-ID: <68b791330905181913m4406de47q8c1bb9396d91038@mail.gmail.com> Actually I think that Bill and James are saying the same thing. Resources are "served up" by virtual servers. There really should be a 1 to 1 relationship: 1 volume --> 1 virtual server. Then, from login scripts or in any kind of a client side binding, you always reference the virtual server (using either edirectory syntax like .vservername.ou.ou: or I think you can also use a unc path referencing the virtual server name but I don't remember now and am not sure that would work with linux. In any event, when you reference the vserver name, it can float (migrate) from host to host and the clients don't know or care which physical host it is on. It is best planning practice to plan/arrange for virtual servers to be spread around your cluster nodes/hosts. So if I have a 3 node cluster I generally put one resource/virtual server on each host. If I have more resources, then I create more virtual servers and some hosts "host" more than one. When a physical host has one vserver, then that host's tools can only "see" that one volume/resource. If that host has two, then it can see two. But it really does not matter which host it is on from the perspective of a non-host client because it attaches using the vserver identification which is independant of the physical node. I've not got person experience with OES clustering, but peers do ... and they are quit clear and explicit that clustering in OES works the way it does in Netware. i think you have gotten yourself bundled somewhere along the line, and a detail or definition has become magnified in your mind, out of proportion -- so now all appears distorted to you. p On Mon, May 18, 2009 at 9:00 PM, joea at j4computers.com wrote: >> >> This is exactly the way it works. ?A user mapped to a virtual resource does >> not have to care. ?All mappings should be done based on virtual cluster >> server eDirectory objects. ?The local resources will not be visible as >> virtual cluster server eDir objects. ?The only way those could be incorrectly >> mapped is if you used server mapped volumes as opposed to eDirectory mapped >> volumes. ?Cluster objects only exists as eDir objects. ?None of the local >> node resources will show up as eDir cluster resources. >> > > That may be the way you see it working on NetWare, but it does not seem to work that way in Linux OES1. > > It works the way I described it, at least on our gear. > > And the way I interpret Bill's response, it differs from what you describe. > > joe a. > > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > From joea at j4computers.com Tue May 19 14:59:53 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 19 May 2009 09:59:53 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A128305020000850005FEFC@FS-LIN-OES> This has gotten a bit, umm, silly. 1 volume = 1 virtual server. I get that. Fine. But that implies, to me, and some of you seem to be saying it works, is that, I can move the volumes (resources), from node to node and still see the volumes by referencing the virtual server. Fine, that's what I want. Seems obvious it should be so. But, frankly, I hear some of you saying this is not what to expect. I'd like us all to be on the same page. Let me lay this out again. I have a two node cluster. N1 and N2. Two resources, R1 and R2. Two VNCP objects VNCP1 and VNCP2. To start - Resource R1 exists on node N1 and is bound to VNCP1. I see it on VNCP1. Resource R2 exists on node N2 and is bound to VNCP2. I see it on VNCP2. Then- If I migrate R1 to node N2, the unload script (running on N1) will unbind VNCP1 from R1's IP, and stop R1. The load script (running on N2) will bind R1's IP to VNCP1 and start R1. The load script says . . .blah . . . ncpcon bind --ncpservername=VNCP1 --ipaddress=R1's_ip_address . . .more blah. . . The unload script says . . .blah . . . ncpcon unbind --ncpservername=VNCP1 --ipaddress=R1's_ip_address . . .more blah. . . Are we still together on this? Anything wrong there? Now, please tell me what I should expect to see in the Virtual Servers. I would expect that VNCP1 should continue to show R1 and the associated volumes, allowing for some "brief" period of non connectivity. However, that is not what happens. Here. VNCP1 is now "empty" and VNCP2 (previously created for R2 and bound to it's IP), now shows both R1 and R2. (I would not care, if it continued to show in VNCP1, as well.) Has anyone done this, successfully, and if so, how? Please, no more examples, analogies, "I've always done it on NetWare", etc., lest I go over the edge, (hmm. . .) just concise response, in line, to what I have laid out. I thank everyone for their tolerance and softspoken-ness. joe a. >>> On 5/18/2009 at 10:13 PM, Peter Van Lone wrote: > Actually I think that Bill and James are saying the same thing. > Resources are "served up" by virtual servers. There really should be a > 1 to 1 relationship: 1 volume --> 1 virtual server. > > Then, from login scripts or in any kind of a client side binding, you > always reference the virtual server (using either edirectory syntax > like .vservername.ou.ou: or I think you can also use a unc path > referencing the virtual server name but I don't remember now and am > not sure that would work with linux. > > In any event, when you reference the vserver name, it can float > (migrate) from host to host and the clients don't know or care which > physical host it is on. It is best planning practice to plan/arrange > for virtual servers to be spread around your cluster nodes/hosts. So > if I have a 3 node cluster I generally put one resource/virtual server > on each host. If I have more resources, then I create more virtual > servers and some hosts "host" more than one. > > When a physical host has one vserver, then that host's tools can only > "see" that one volume/resource. If that host has two, then it can see > two. But it really does not matter which host it is on from the > perspective of a non-host client because it attaches using the vserver > identification which is independant of the physical node. > > I've not got person experience with OES clustering, but peers do ... > and they are quit clear and explicit that clustering in OES works the > way it does in Netware. i think you have gotten yourself bundled > somewhere along the line, and a detail or definition has become > magnified in your mind, out of proportion -- so now all appears > distorted to you. > > p > > > > On Mon, May 18, 2009 at 9:00 PM, joea at j4computers.com > wrote: >>> >>> This is exactly the way it works. A user mapped to a virtual resource does >>> not have to care. All mappings should be done based on virtual cluster >>> server eDirectory objects. The local resources will not be visible as >>> virtual cluster server eDir objects. The only way those could be > incorrectly >>> mapped is if you used server mapped volumes as opposed to eDirectory mapped >>> volumes. Cluster objects only exists as eDir objects. None of the local >>> node resources will show up as eDir cluster resources. >>> >> >> That may be the way you see it working on NetWare, but it does not seem to > work that way in Linux OES1. >> >> It works the way I described it, at least on our gear. >> >> And the way I interpret Bill's response, it differs from what you describe. >> >> joe a. >> >> >> >> _______________________________________________ >> Novell mailing list >> Novell at netlab1.oucs.ox.ac.uk >> http://netlab1.usu.edu/mailman/listinfo/novell >> > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell From bbrush at gmail.com Tue May 19 15:08:03 2009 From: bbrush at gmail.com (Bill Brush) Date: Tue, 19 May 2009 09:08:03 -0500 Subject: Map to cluster volume on Linux (ncp) In-Reply-To: <4A128305020000850005FEFC@FS-LIN-OES> References: <4A128305020000850005FEFC@FS-LIN-OES> Message-ID: <167f4090905190708x7df62b49u1c69faba2d38930a@mail.gmail.com> On Tue, May 19, 2009 at 8:59 AM, joea at j4computers.com wrote: > This has gotten a bit, umm, silly. > > > > Now, please tell me what I should expect to see in the Virtual Servers. I > would > expect that VNCP1 should continue to show R1 and the associated volumes, > allowing for some "brief" period of non connectivity. > > However, that is not what happens. Here. VNCP1 is now "empty" and VNCP2 > (previously created for R2 and bound to it's IP), now shows both R1 and R2. > (I would > not care, if it continued to show in VNCP1, as well.) > > Has anyone done this, successfully, and if so, how? Please, no more > examples, analogies, > "I've always done it on NetWare", etc., lest I go over the edge, (hmm. . > .) just concise > response, in line, to what I have laid out. > > I thank everyone for their tolerance and softspoken-ness. > If one physical node is hosting both virtual servers, and both cluster volumes, then both virtual servers should show BOTH cluster volumes. If I have it right then both R1 and R2 should be visible under VNCP1 and VNCP2 (as well as N2.) If this is not what's visible then you do indeed have an issue. Bill From joea at j4computers.com Tue May 19 16:54:11 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 19 May 2009 11:54:11 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A129DD2020000850005FF11@FS-LIN-OES> Sigh. The LGM's are out in force, this week. I was off attending to other matters, leaving one VNCP1 empty and VNCP2 with both volumes visible. Several hours later, I take another look and the volumes appear where expected. I know patience is a virtue, but . . . how long is reasonable? In past attempts, I have rechecked every few minutes for up to 15 minutes, without seeing changes. SLP issues? It appears, at first glance, that the resources have not been migrated or otherwise toyed with and everything is just as I left it. I'm not sure if I am pleased, or troubled. Some would say the latter, regardless of context. Oh, Nurse . . . joe a. >> If one physical node is hosting both virtual servers, and both cluster volumes, then both virtual servers should show BOTH cluster volumes. If I have it right then both R1 and R2 should be visible under VNCP1 and VNCP2 (as well as N2.) If this is not what's visible then you do indeed have an issue. Bill _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Tue May 19 16:58:36 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Tue, 19 May 2009 11:58:36 -0400 Subject: Map to cluster volume on Linux (ncp) Message-ID: <4A129EDC020000850005FF18@FS-LIN-OES> Please ignore the message. It is all a lie. It is just as I left it, both volumes under one VNCP object. I apologize for the false alarm. My bad. joe a. >>> "joea at j4computers.com" 05/19/09 11:55 AM >>> Sigh. The LGM's are out in force, this week. I was off attending to other matters, leaving one VNCP1 empty and VNCP2 with both volumes visible. Several hours later, I take another look and the volumes appear where expected. I know patience is a virtue, but . . . how long is reasonable? In past attempts, I have rechecked every few minutes for up to 15 minutes, without seeing changes. SLP issues? It appears, at first glance, that the resources have not been migrated or otherwise toyed with and everything is just as I left it. I'm not sure if I am pleased, or troubled. Some would say the latter, regardless of context. Oh, Nurse . . . joe a. >> If one physical node is hosting both virtual servers, and both cluster volumes, then both virtual servers should show BOTH cluster volumes. If I have it right then both R1 and R2 should be visible under VNCP1 and VNCP2 (as well as N2.) If this is not what's visible then you do indeed have an issue. Bill _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From zzz at minneapolis.edu Thu May 21 01:24:34 2009 From: zzz at minneapolis.edu (zzz) Date: Wed, 20 May 2009 19:24:34 -0500 Subject: ABENDS when backing up with COMMVAULT Message-ID: Greetings, NW6.0 SP5 and TFS volumes for SYS and DATA with NFS on both. We upgraded from COMMVAULT 6.1 to CV 7.0 about a week ago. Things were working until a day or so ago. Now, THIS server abends whenever we start the backup job. All other servers still backing up fine. I have been working with CV support on this, but of course, heard the typical thing that they had never seen this before. Clearly, UNIXMoveExtendedName and the fact that NFS.NAM is referenced in the abend log points to something with the NFS name space that the CV GALAXY agents are not liking. Maybe corruption or something? Still have not had time to isolate to the DATA or the SYS volume yet. Of course, the DATA volume on this box is 250 GIG, so any move or restore would take time. The DATA volume has NFS but I don't think it is needed. However, removal of NFS could be risky as well. We do have MAC OSX, but I think all they need is MAC name space. ABEND log clip is below and open to any suggestions. Thank you Dana ********************************************************* Server EARTH halted Tuesday, May 19, 2009 4:53:15.554 pm Abend 1 on P00: Server-5.60.05-0: UNIXMoveExtendedName wrong extant number. Registers: CS = 0008 DS = 0010 ES = 0010 FS = 0010 GS = 0010 SS = 0010 EAX = 8AB1B000 EBX = CE26D280 ECX = 00000000 EDX = 00000001 ESI = A461B068 EDI = CE26D280 EBP = A461AD7A ESP = A461ACEC EIP = 8AAF7795 FLAGS = 00000246 8AAF7795 83C404 ADD ESP,00000004 EIP in NFS.NAM at code start +00003795h The violation occurred while processing the following instruction: 8AAF7795 83C404 ADD ESP,00000004 8AAF7798 807B3500 CMP [EBX+35],00 8AAF779C 7513 JNZ 8AAF77B1 8AAF779E A1B8289825 MOV EAX,[259828B8]=8AB1B000 8AAF77A3 FFB0D4010000 PUSH dword ptr [EAX+000001D4] 8AAF77A9 E838926175 CALL SERVER.NLM|Abend 8AAF77AE 83C404 ADD ESP,00000004 8AAF77B1 8D8576FFFFFF LEA EAX,[EBP-0000008A] 8AAF77B7 50 PUSH EAX 8AAF77B8 0FB64335 MOVZX EAX,byte ptr [EBX+35] Running process: CVFSScanMain 1 Process Thread Owned by NLM: FSIFIND.NLM Stack pointer: A461C7C8 OS Stack limit: A45FDB60 Scheduling priority: 67371008 Wait state: 5050100 Delayed Stack: --8AB1C1E7 ? --00000000 ? --00000000 ? CB67AD9B (FILESYS.NLM|INWGetEntryFromPathStringBase+16B) --0000026A ? --820EAB80 ? --FFFF0000 ? --CE26D100 ? --CE26D100 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --0002D97C (LOADER.EXE|startPublicList+A0D4) --40000010 ? --00062294 ? --00000000 ? --00000000 ? --0000FFFF ? --00000015 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000002 ? --00000002 ? --0002D97E (LOADER.EXE|startPublicList+A0D6) --00000001 ? CB67B27C (FILESYS.NLM|INWGetOtherNameSpaceEntry+1C8) CB67B325 (FILESYS.NLM|INWGetOtherNameSpaceEntry+271) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D200 ? --00000000 ? --00000008 ? --A461AF18 ? --CE26D100 ? --AD02D97C ? --00000002 ? CBA0A7A5 (NWSA.NSS|ZH_GetOtherNameSpaceEntry+165) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D100 ? --00000000 ? --00000001 ? --A461AF3C ? --A461AE08 ? --CE26D280 ? 8AAF50E5 (NFS.NAM|SetWholeNFSBitMap+F71) --00000002 ? --CE26D280 ? --A461B091 ? --00000002 ? --A461B068 ? --A461B16C ? CB68A1FA (FILESYS.NLM|GenericFillReplyBuffer+30E) --00000002 ? --CE26D280 ? --A461B068 ? --00000002 ? --0002D97F (LOADER.EXE|startPublicList+A0D7) --00000002 ? --00000000 ? CB67B1CB (FILESYS.NLM|INWGetOtherNameSpaceEntry+117) CB67B258 (FILESYS.NLM|INWGetOtherNameSpaceEntry+1A4) --A461AE94 ? --00000000 ? --00000002 ? --00000002 ? CB6877C7 (FILESYS.NLM|INWF3ObtainFileInfo+24F) --00000002 ? --0000026A ? --AD02D97F ? --AD02D97C ? --00000002 ? --00000001 ? --CE26D280 ? --CE26D100 ? --A461B16C ? --A461B068 ? --A461AEDC ? Additional Information: The NetWare OS detected a problem with the system while executing a process owned by NFS.NAM. It may be the source of the problem or there may have been a memory corruption. Loaded Modules: FSIFIND.NLM Galaxy for NetWare File System Scanner Version 6.10 November 6, 2006 Code Address: 91B8D000h Length: 00003266h Data Address: 9180B000h Length: 00002104h EVMGRC.NLM Galaxy for NetWare Event Manager Client Version 6.10 March 26, 2007 Code Address: 8F86F000h Length: 00009EA2h Data Address: 8F879000h Length: 000053B4h CVD.NLM Galaxy for NetWare Communications Service Version 6.10 March 26, 2007 Code Address: 8F7DC000h Length: 0001C63Fh Data Address: 8F7F9000h Length: 0000ADD4h NDSBRWSE.NLM Galaxy for NetWare NDS Browser Version 6.10 March 16, 2006 Code Address: 8F77F000h Length: 000026CEh Data Address: 8F782000h Length: 00000F7Ch FSBRWSE.NLM Galaxy for NetWare File System Browser Version 6.10 March 26, 2007 Code Address: 8F752000h Length: 0000278Eh Data Address: 8F757000h Length: 0000113Ch CVSMS.NLM Galaxy for NetWare SMS Interface Version 6.10 July 12, 2007 Code Address: 8F716000h Length: 0000711Dh Data Address: 8F722000h Length: 00005C9Ch CLBROWSE.NLM Galaxy for NetWare Client Browse Version 6.10 March 26, 2007 Code Address: 8F6CF000h Length: 000033EEh Data Address: 8F6DF000h Length: 000066ACh CLRESTOR.NLM Galaxy for NetWare Client Restore Version 6.10 March 26, 2007 Code Address: 8F69B000h Length: 00010B24h Data Address: 8F6B4000h Length: 00007B3Ch CLBACKUP.NLM Galaxy for NetWare Client Backup Version 6.10 June 5, 2007 Code Address: 8F641000h Length: 00014754h Data Address: 8F616000h Length: 0000B61Ch CVNETCHK.NLM Galaxy for NetWare Network Check Version 6.10 November 17, 2005 Code Address: 8F608000h Length: 00003B7Ah Data Address: 8F5AC000h Length: 00001924h CVARCH.NLM Galaxy for NetWare Archive Library Version 6.10 March 29, 2007 Code Address: 8F59F000h Length: 0000CB9Ah Data Address: 8F56E000h Length: 00006A0Ch PROCMODS.NLM Galaxy for NetWare PipeLine Modules Version 6.10 November 17, 2005 Code Address: 8F53B000h Length: 000098BFh Data Address: 8F45E000h Length: 0000585Ch CVCRYPT.NLM Galaxy for NetWare Cryptography Library Version 6.10 March 26, 2007 Code Address: 8F50D000h Length: 00021178h Data Address: 8F423000h Length: 000091D5h CVLZOLIB.NLM Galaxy for NetWare LZO Compression Library Version 6.10 November 17, 2005 Code Address: 8F40C000h Length: 00001D1Dh Data Address: 8F40E000h Length: 00000748h CVZLIB.NLM Galaxy for NetWare ZIP Compression Library Version 6.10 November 17, 2005 Code Address: 8F3AC000h Length: 000088BDh Data Address: 8F2BB000h Length: 000048E4h CVJOBCL.NLM Galaxy for NetWare Job Client Version 6.10 August 22, 2006 Code Address: 8F362000h Length: 0002A32Fh Data Address: 8F05B000h Length: 00018564h CVSIM.NLM Galaxy for NetWare Software Installation Manager Version 6.10 November 17, 2005 Code Address: 8F03A000h Length: 0000426Dh Data Address: 8F03F000h Length: 00001CACh CVAPPMGR.NLM Galaxy for NetWare AppManager Version 6.10 November 17, 2005 Code Address: A50C8000h Length: 00090B74h Data Address: 8F3B6000h Length: 0004817Ch CVLIB.NLM CommVault Library for NetWare Version 6.10 August 9, 2007 Code Address: 8F2C2000h Length: 0009E0BAh Data Address: 8EF95000h Length: 0003B486h GALAXY.NLM Galaxy for NetWare Loader Version 6.10 November 17, 2005 Code Address: 8E2C9000h Length: 00003A0Dh Data Address: 8E2DC000h Length: 00002EAEh JNDPS.NLM Native Wrapper Java Class Libraries for NDPS From dtran at ssc.ucla.edu Thu May 21 15:37:20 2009 From: dtran at ssc.ucla.edu (Daniel Tran) Date: Thu, 21 May 2009 07:37:20 -0700 Subject: ABENDS when backing up with COMMVAULT In-Reply-To: References: Message-ID: Dana, I clearly don't know why you're getting that abend. Even though your commserve has been updated to 7.0 (simpana) but your NLMs on the server is still 6.1. Could this be the problem? Did you install commvault galaxy sp4 for netware? FWIW, I have a similar configuration here with nw6.0sp5 without nfs namespace I also have stuff from the 7.0 installed on the another server. Commvault 7.0 is not supported on netware 6 (that's what they said) but you can work around to get it installed. Not sure if this is something you want to try. If you do then you have to do the following. - backup your sys:/galaxy folder (make copy elsewhere) - install cv 7 from CD to netware server, it will detect as upgrade from 6.1 to 7 - when done, galaxy will fail to load because sys:system\galaxy.lod has netware system = 5.70. If you change that to 5.60 with an editor, it will load. On the server, I have the following patches installed in this order (post nw6sp5): - tsaup21 - mm60sp5 - slp123 - wsock6L - nwlib6j - libcsp6a - nwlib6L Daniel -----Original Message----- From: novell-bounces at netlab1.oucs.ox.ac.uk [mailto:novell-bounces at netlab1.oucs.ox.ac.uk] On Behalf Of zzz Sent: Wednesday, May 20, 2009 5:25 PM To: novell at netlab1.oucs.ox.ac.uk Subject: ABENDS when backing up with COMMVAULT Greetings, NW6.0 SP5 and TFS volumes for SYS and DATA with NFS on both. We upgraded from COMMVAULT 6.1 to CV 7.0 about a week ago. Things were working until a day or so ago. Now, THIS server abends whenever we start the backup job. All other servers still backing up fine. I have been working with CV support on this, but of course, heard the typical thing that they had never seen this before. Clearly, UNIXMoveExtendedName and the fact that NFS.NAM is referenced in the abend log points to something with the NFS name space that the CV GALAXY agents are not liking. Maybe corruption or something? Still have not had time to isolate to the DATA or the SYS volume yet. Of course, the DATA volume on this box is 250 GIG, so any move or restore would take time. The DATA volume has NFS but I don't think it is needed. However, removal of NFS could be risky as well. We do have MAC OSX, but I think all they need is MAC name space. ABEND log clip is below and open to any suggestions. Thank you Dana ********************************************************* Server EARTH halted Tuesday, May 19, 2009 4:53:15.554 pm Abend 1 on P00: Server-5.60.05-0: UNIXMoveExtendedName wrong extant number. Registers: CS = 0008 DS = 0010 ES = 0010 FS = 0010 GS = 0010 SS = 0010 EAX = 8AB1B000 EBX = CE26D280 ECX = 00000000 EDX = 00000001 ESI = A461B068 EDI = CE26D280 EBP = A461AD7A ESP = A461ACEC EIP = 8AAF7795 FLAGS = 00000246 8AAF7795 83C404 ADD ESP,00000004 EIP in NFS.NAM at code start +00003795h The violation occurred while processing the following instruction: 8AAF7795 83C404 ADD ESP,00000004 8AAF7798 807B3500 CMP [EBX+35],00 8AAF779C 7513 JNZ 8AAF77B1 8AAF779E A1B8289825 MOV EAX,[259828B8]=8AB1B000 8AAF77A3 FFB0D4010000 PUSH dword ptr [EAX+000001D4] 8AAF77A9 E838926175 CALL SERVER.NLM|Abend 8AAF77AE 83C404 ADD ESP,00000004 8AAF77B1 8D8576FFFFFF LEA EAX,[EBP-0000008A] 8AAF77B7 50 PUSH EAX 8AAF77B8 0FB64335 MOVZX EAX,byte ptr [EBX+35] Running process: CVFSScanMain 1 Process Thread Owned by NLM: FSIFIND.NLM Stack pointer: A461C7C8 OS Stack limit: A45FDB60 Scheduling priority: 67371008 Wait state: 5050100 Delayed Stack: --8AB1C1E7 ? --00000000 ? --00000000 ? CB67AD9B (FILESYS.NLM|INWGetEntryFromPathStringBase+16B) --0000026A ? --820EAB80 ? --FFFF0000 ? --CE26D100 ? --CE26D100 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --0002D97C (LOADER.EXE|startPublicList+A0D4) --40000010 ? --00062294 ? --00000000 ? --00000000 ? --0000FFFF ? --00000015 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000002 ? --00000002 ? --0002D97E (LOADER.EXE|startPublicList+A0D6) --00000001 ? CB67B27C (FILESYS.NLM|INWGetOtherNameSpaceEntry+1C8) CB67B325 (FILESYS.NLM|INWGetOtherNameSpaceEntry+271) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D200 ? --00000000 ? --00000008 ? --A461AF18 ? --CE26D100 ? --AD02D97C ? --00000002 ? CBA0A7A5 (NWSA.NSS|ZH_GetOtherNameSpaceEntry+165) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D100 ? --00000000 ? --00000001 ? --A461AF3C ? --A461AE08 ? --CE26D280 ? 8AAF50E5 (NFS.NAM|SetWholeNFSBitMap+F71) --00000002 ? --CE26D280 ? --A461B091 ? --00000002 ? --A461B068 ? --A461B16C ? CB68A1FA (FILESYS.NLM|GenericFillReplyBuffer+30E) --00000002 ? --CE26D280 ? --A461B068 ? --00000002 ? --0002D97F (LOADER.EXE|startPublicList+A0D7) --00000002 ? --00000000 ? CB67B1CB (FILESYS.NLM|INWGetOtherNameSpaceEntry+117) CB67B258 (FILESYS.NLM|INWGetOtherNameSpaceEntry+1A4) --A461AE94 ? --00000000 ? --00000002 ? --00000002 ? CB6877C7 (FILESYS.NLM|INWF3ObtainFileInfo+24F) --00000002 ? --0000026A ? --AD02D97F ? --AD02D97C ? --00000002 ? --00000001 ? --CE26D280 ? --CE26D100 ? --A461B16C ? --A461B068 ? --A461AEDC ? Additional Information: The NetWare OS detected a problem with the system while executing a process owned by NFS.NAM. It may be the source of the problem or there may have been a memory corruption. Loaded Modules: FSIFIND.NLM Galaxy for NetWare File System Scanner Version 6.10 November 6, 2006 Code Address: 91B8D000h Length: 00003266h Data Address: 9180B000h Length: 00002104h EVMGRC.NLM Galaxy for NetWare Event Manager Client Version 6.10 March 26, 2007 Code Address: 8F86F000h Length: 00009EA2h Data Address: 8F879000h Length: 000053B4h CVD.NLM Galaxy for NetWare Communications Service Version 6.10 March 26, 2007 Code Address: 8F7DC000h Length: 0001C63Fh Data Address: 8F7F9000h Length: 0000ADD4h NDSBRWSE.NLM Galaxy for NetWare NDS Browser Version 6.10 March 16, 2006 Code Address: 8F77F000h Length: 000026CEh Data Address: 8F782000h Length: 00000F7Ch FSBRWSE.NLM Galaxy for NetWare File System Browser Version 6.10 March 26, 2007 Code Address: 8F752000h Length: 0000278Eh Data Address: 8F757000h Length: 0000113Ch CVSMS.NLM Galaxy for NetWare SMS Interface Version 6.10 July 12, 2007 Code Address: 8F716000h Length: 0000711Dh Data Address: 8F722000h Length: 00005C9Ch CLBROWSE.NLM Galaxy for NetWare Client Browse Version 6.10 March 26, 2007 Code Address: 8F6CF000h Length: 000033EEh Data Address: 8F6DF000h Length: 000066ACh CLRESTOR.NLM Galaxy for NetWare Client Restore Version 6.10 March 26, 2007 Code Address: 8F69B000h Length: 00010B24h Data Address: 8F6B4000h Length: 00007B3Ch CLBACKUP.NLM Galaxy for NetWare Client Backup Version 6.10 June 5, 2007 Code Address: 8F641000h Length: 00014754h Data Address: 8F616000h Length: 0000B61Ch CVNETCHK.NLM Galaxy for NetWare Network Check Version 6.10 November 17, 2005 Code Address: 8F608000h Length: 00003B7Ah Data Address: 8F5AC000h Length: 00001924h CVARCH.NLM Galaxy for NetWare Archive Library Version 6.10 March 29, 2007 Code Address: 8F59F000h Length: 0000CB9Ah Data Address: 8F56E000h Length: 00006A0Ch PROCMODS.NLM Galaxy for NetWare PipeLine Modules Version 6.10 November 17, 2005 Code Address: 8F53B000h Length: 000098BFh Data Address: 8F45E000h Length: 0000585Ch CVCRYPT.NLM Galaxy for NetWare Cryptography Library Version 6.10 March 26, 2007 Code Address: 8F50D000h Length: 00021178h Data Address: 8F423000h Length: 000091D5h CVLZOLIB.NLM Galaxy for NetWare LZO Compression Library Version 6.10 November 17, 2005 Code Address: 8F40C000h Length: 00001D1Dh Data Address: 8F40E000h Length: 00000748h CVZLIB.NLM Galaxy for NetWare ZIP Compression Library Version 6.10 November 17, 2005 Code Address: 8F3AC000h Length: 000088BDh Data Address: 8F2BB000h Length: 000048E4h CVJOBCL.NLM Galaxy for NetWare Job Client Version 6.10 August 22, 2006 Code Address: 8F362000h Length: 0002A32Fh Data Address: 8F05B000h Length: 00018564h CVSIM.NLM Galaxy for NetWare Software Installation Manager Version 6.10 November 17, 2005 Code Address: 8F03A000h Length: 0000426Dh Data Address: 8F03F000h Length: 00001CACh CVAPPMGR.NLM Galaxy for NetWare AppManager Version 6.10 November 17, 2005 Code Address: A50C8000h Length: 00090B74h Data Address: 8F3B6000h Length: 0004817Ch CVLIB.NLM CommVault Library for NetWare Version 6.10 August 9, 2007 Code Address: 8F2C2000h Length: 0009E0BAh Data Address: 8EF95000h Length: 0003B486h GALAXY.NLM Galaxy for NetWare Loader Version 6.10 November 17, 2005 Code Address: 8E2C9000h Length: 00003A0Dh Data Address: 8E2DC000h Length: 00002EAEh JNDPS.NLM Native Wrapper Java Class Libraries for NDPS _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From zzz at minneapolis.edu Fri May 22 02:01:43 2009 From: zzz at minneapolis.edu (zzz) Date: Thu, 21 May 2009 20:01:43 -0500 Subject: ABENDS COMMVAULT (DATA CORRUPTION) Message-ID: Hi Daniel, Thanks for the post. CV support went through and looked at everything, and we finally reinstalled the 6.1 client just to rule that out. Next I broke SYS and DATA into two separate jobs to see which volume was crashing. Unfortunately, SYS backs up just fine - the 250 GIG DATA HOMEDIR volume is the one with the NFS corruption issue. And my last backup is Sunday. Remember, users can still access the data - we just can't back it up. This is SAN attached, so I am thinking of creating another volume WITHOUT NFS, and using something (suggestions?) to copy EXISTING volume to NEW volume, then see if I can backup the NEW volume. If that works, I may be able to restore rights from the sunday CV backup, but will have to think about this. Open to all thoughts and I will think about this tonight. Thanks for the help. Dana >>> dtran at ssc.ucla.edu 5/21/2009 9:37:20 AM >>> Dana, I clearly don't know why you're getting that abend. Even though your commserve has been updated to 7.0 (simpana) but your NLMs on the server is still 6.1. Could this be the problem? Did you install commvault galaxy sp4 for netware? FWIW, I have a similar configuration here with nw6.0sp5 without nfs namespace I also have stuff from the 7.0 installed on the another server. Commvault 7.0 is not supported on netware 6 (that's what they said) but you can work around to get it installed. Not sure if this is something you want to try. If you do then you have to do the following. - backup your sys:/galaxy folder (make copy elsewhere) - install cv 7 from CD to netware server, it will detect as upgrade from 6.1 to 7 - when done, galaxy will fail to load because sys:system\galaxy.lod has netware system = 5.70. If you change that to 5.60 with an editor, it will load. On the server, I have the following patches installed in this order (post nw6sp5): - tsaup21 - mm60sp5 - slp123 - wsock6L - nwlib6j - libcsp6a - nwlib6L Daniel -----Original Message----- From: novell-bounces at netlab1.oucs.ox.ac.uk [mailto:novell-bounces at netlab1.oucs.ox.ac.uk] On Behalf Of zzz Sent: Wednesday, May 20, 2009 5:25 PM To: novell at netlab1.oucs.ox.ac.uk Subject: ABENDS when backing up with COMMVAULT Greetings, NW6.0 SP5 and TFS volumes for SYS and DATA with NFS on both. We upgraded from COMMVAULT 6.1 to CV 7.0 about a week ago. Things were working until a day or so ago. Now, THIS server abends whenever we start the backup job. All other servers still backing up fine. I have been working with CV support on this, but of course, heard the typical thing that they had never seen this before. Clearly, UNIXMoveExtendedName and the fact that NFS.NAM is referenced in the abend log points to something with the NFS name space that the CV GALAXY agents are not liking. Maybe corruption or something? Still have not had time to isolate to the DATA or the SYS volume yet. Of course, the DATA volume on this box is 250 GIG, so any move or restore would take time. The DATA volume has NFS but I don't think it is needed. However, removal of NFS could be risky as well. We do have MAC OSX, but I think all they need is MAC name space. ABEND log clip is below and open to any suggestions. Thank you Dana ********************************************************* Server EARTH halted Tuesday, May 19, 2009 4:53:15.554 pm Abend 1 on P00: Server-5.60.05-0: UNIXMoveExtendedName wrong extant number. Registers: CS = 0008 DS = 0010 ES = 0010 FS = 0010 GS = 0010 SS = 0010 EAX = 8AB1B000 EBX = CE26D280 ECX = 00000000 EDX = 00000001 ESI = A461B068 EDI = CE26D280 EBP = A461AD7A ESP = A461ACEC EIP = 8AAF7795 FLAGS = 00000246 8AAF7795 83C404 ADD ESP,00000004 EIP in NFS.NAM at code start +00003795h The violation occurred while processing the following instruction: 8AAF7795 83C404 ADD ESP,00000004 8AAF7798 807B3500 CMP [EBX+35],00 8AAF779C 7513 JNZ 8AAF77B1 8AAF779E A1B8289825 MOV EAX,[259828B8]=8AB1B000 8AAF77A3 FFB0D4010000 PUSH dword ptr [EAX+000001D4] 8AAF77A9 E838926175 CALL SERVER.NLM|Abend 8AAF77AE 83C404 ADD ESP,00000004 8AAF77B1 8D8576FFFFFF LEA EAX,[EBP-0000008A] 8AAF77B7 50 PUSH EAX 8AAF77B8 0FB64335 MOVZX EAX,byte ptr [EBX+35] Running process: CVFSScanMain 1 Process Thread Owned by NLM: FSIFIND.NLM Stack pointer: A461C7C8 OS Stack limit: A45FDB60 Scheduling priority: 67371008 Wait state: 5050100 Delayed Stack: --8AB1C1E7 ? --00000000 ? --00000000 ? CB67AD9B (FILESYS.NLM|INWGetEntryFromPathStringBase+16B) --0000026A ? --820EAB80 ? --FFFF0000 ? --CE26D100 ? --CE26D100 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --0002D97C (LOADER.EXE|startPublicList+A0D4) --40000010 ? --00062294 ? --00000000 ? --00000000 ? --0000FFFF ? --00000015 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000000 ? --00000002 ? --00000002 ? --0002D97E (LOADER.EXE|startPublicList+A0D6) --00000001 ? CB67B27C (FILESYS.NLM|INWGetOtherNameSpaceEntry+1C8) CB67B325 (FILESYS.NLM|INWGetOtherNameSpaceEntry+271) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D200 ? --00000000 ? --00000008 ? --A461AF18 ? --CE26D100 ? --AD02D97C ? --00000002 ? CBA0A7A5 (NWSA.NSS|ZH_GetOtherNameSpaceEntry+165) --00000002 ? --0002D97C (LOADER.EXE|startPublicList+A0D4) --CE26D100 ? --00000000 ? --00000001 ? --A461AF3C ? --A461AE08 ? --CE26D280 ? 8AAF50E5 (NFS.NAM|SetWholeNFSBitMap+F71) --00000002 ? --CE26D280 ? --A461B091 ? --00000002 ? --A461B068 ? --A461B16C ? CB68A1FA (FILESYS.NLM|GenericFillReplyBuffer+30E) --00000002 ? --CE26D280 ? --A461B068 ? --00000002 ? --0002D97F (LOADER.EXE|startPublicList+A0D7) --00000002 ? --00000000 ? CB67B1CB (FILESYS.NLM|INWGetOtherNameSpaceEntry+117) CB67B258 (FILESYS.NLM|INWGetOtherNameSpaceEntry+1A4) --A461AE94 ? --00000000 ? --00000002 ? --00000002 ? CB6877C7 (FILESYS.NLM|INWF3ObtainFileInfo+24F) --00000002 ? --0000026A ? --AD02D97F ? --AD02D97C ? --00000002 ? --00000001 ? --CE26D280 ? --CE26D100 ? --A461B16C ? --A461B068 ? --A461AEDC ? Additional Information: The NetWare OS detected a problem with the system while executing a process owned by NFS.NAM. It may be the source of the problem or there may have been a memory corruption. Loaded Modules: FSIFIND.NLM Galaxy for NetWare File System Scanner Version 6.10 November 6, 2006 Code Address: 91B8D000h Length: 00003266h Data Address: 9180B000h Length: 00002104h EVMGRC.NLM Galaxy for NetWare Event Manager Client Version 6.10 March 26, 2007 Code Address: 8F86F000h Length: 00009EA2h Data Address: 8F879000h Length: 000053B4h CVD.NLM Galaxy for NetWare Communications Service Version 6.10 March 26, 2007 Code Address: 8F7DC000h Length: 0001C63Fh Data Address: 8F7F9000h Length: 0000ADD4h NDSBRWSE.NLM Galaxy for NetWare NDS Browser Version 6.10 March 16, 2006 Code Address: 8F77F000h Length: 000026CEh Data Address: 8F782000h Length: 00000F7Ch FSBRWSE.NLM Galaxy for NetWare File System Browser Version 6.10 March 26, 2007 Code Address: 8F752000h Length: 0000278Eh Data Address: 8F757000h Length: 0000113Ch CVSMS.NLM Galaxy for NetWare SMS Interface Version 6.10 July 12, 2007 Code Address: 8F716000h Length: 0000711Dh Data Address: 8F722000h Length: 00005C9Ch CLBROWSE.NLM Galaxy for NetWare Client Browse Version 6.10 March 26, 2007 Code Address: 8F6CF000h Length: 000033EEh Data Address: 8F6DF000h Length: 000066ACh CLRESTOR.NLM Galaxy for NetWare Client Restore Version 6.10 March 26, 2007 Code Address: 8F69B000h Length: 00010B24h Data Address: 8F6B4000h Length: 00007B3Ch CLBACKUP.NLM Galaxy for NetWare Client Backup Version 6.10 June 5, 2007 Code Address: 8F641000h Length: 00014754h Data Address: 8F616000h Length: 0000B61Ch CVNETCHK.NLM Galaxy for NetWare Network Check Version 6.10 November 17, 2005 Code Address: 8F608000h Length: 00003B7Ah Data Address: 8F5AC000h Length: 00001924h CVARCH.NLM Galaxy for NetWare Archive Library Version 6.10 March 29, 2007 Code Address: 8F59F000h Length: 0000CB9Ah Data Address: 8F56E000h Length: 00006A0Ch PROCMODS.NLM Galaxy for NetWare PipeLine Modules Version 6.10 November 17, 2005 Code Address: 8F53B000h Length: 000098BFh Data Address: 8F45E000h Length: 0000585Ch CVCRYPT.NLM Galaxy for NetWare Cryptography Library Version 6.10 March 26, 2007 Code Address: 8F50D000h Length: 00021178h Data Address: 8F423000h Length: 000091D5h CVLZOLIB.NLM Galaxy for NetWare LZO Compression Library Version 6.10 November 17, 2005 Code Address: 8F40C000h Length: 00001D1Dh Data Address: 8F40E000h Length: 00000748h CVZLIB.NLM Galaxy for NetWare ZIP Compression Library Version 6.10 November 17, 2005 Code Address: 8F3AC000h Length: 000088BDh Data Address: 8F2BB000h Length: 000048E4h CVJOBCL.NLM Galaxy for NetWare Job Client Version 6.10 August 22, 2006 Code Address: 8F362000h Length: 0002A32Fh Data Address: 8F05B000h Length: 00018564h CVSIM.NLM Galaxy for NetWare Software Installation Manager Version 6.10 November 17, 2005 Code Address: 8F03A000h Length: 0000426Dh Data Address: 8F03F000h Length: 00001CACh CVAPPMGR.NLM Galaxy for NetWare AppManager Version 6.10 November 17, 2005 Code Address: A50C8000h Length: 00090B74h Data Address: 8F3B6000h Length: 0004817Ch CVLIB.NLM CommVault Library for NetWare Version 6.10 August 9, 2007 Code Address: 8F2C2000h Length: 0009E0BAh Data Address: 8EF95000h Length: 0003B486h GALAXY.NLM Galaxy for NetWare Loader Version 6.10 November 17, 2005 Code Address: 8E2C9000h Length: 00003A0Dh Data Address: 8E2DC000h Length: 00002EAEh JNDPS.NLM Native Wrapper Java Class Libraries for NDPS _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From James.Taylor at eastcobbgroup.com Fri May 22 04:00:22 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 21 May 2009 23:00:22 -0400 Subject: ABENDS COMMVAULT (DATA CORRUPTION) In-Reply-To: References: Message-ID: <4A15DD0602000075000326B3@inet.eastcobbgroup.com> You can use the SCMT (Server Consolidation and Migration Tool) to move the data to a new volume. It is very fast and will retain the trustees for the files. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "zzz" 5/21/2009 09:01 PM >>> This is SAN attached, so I am thinking of creating another volume WITHOUT NFS, and using something (suggestions?) to copy EXISTING volume to NEW volume, then see if I can backup the NEW volume. If that works, I may be able to restore rights from the sunday CV backup, From bfeder at timeequities.com Tue May 26 15:06:56 2009 From: bfeder at timeequities.com (Bruce Feder) Date: Tue, 26 May 2009 10:06:56 -0400 Subject: OES2 and HP SAN Message-ID: I have been using netware for years and am starting a migration to OES2. Has anyone had any experience connecting and OES2 server to an HP EVA 4000 SAN? Thanks ______________ Bruce Feder MIS Director Time Equities, Inc. bfeder at timeequities.com (P) 212-206-6106 (F) 212-627-9279 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: Bruce Feder.vcf URL: From ksebolt at franciscan.edu Tue May 26 15:03:19 2009 From: ksebolt at franciscan.edu (Kevin Sebolt) Date: Tue, 26 May 2009 10:03:19 -0400 Subject: OES2 and HP SAN (Out of the office) Message-ID: I will be out of the office the week of May 25, 2009. I will respond to your e-mail the week of June 1. Thank you. From peschmid at mpsomaha.org Tue May 26 19:13:30 2009 From: peschmid at mpsomaha.org (Patrick Schmidt) Date: Tue, 26 May 2009 13:13:30 -0500 Subject: ZCM 10.2? Message-ID: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> Does anyone have any insight on SP2 for ZCM will be released? -------------------------------------------------- Patrick Schmidt Millard Public Schools Network Support Specialist (402)715-6278 peschmid at mpsomaha.org --------------------------------------------------- From PHasenjager at kcumb.edu Tue May 26 19:27:13 2009 From: PHasenjager at kcumb.edu (Patrick Hasenjager) Date: Tue, 26 May 2009 13:27:13 -0500 Subject: ZCM 10.2? In-Reply-To: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> Message-ID: <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> Last I heard was either today or tomorrow... That was on Friday. Patrick A. Hasenjager Network Administrator Kansas City University of Medicine and Biosciences phone 816.283.2478 fax 816.283.0692 email phasenjager at kcumb.edu >>> On 5/26/2009 at 1:13 PM, "Patrick Schmidt" wrote: Does anyone have any insight on SP2 for ZCM will be released? -------------------------------------------------- Patrick Schmidt Millard Public Schools Network Support Specialist (402)715-6278 peschmid at mpsomaha.org --------------------------------------------------- _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From PHasenjager at kcumb.edu Tue May 26 19:27:13 2009 From: PHasenjager at kcumb.edu (Patrick Hasenjager) Date: Tue, 26 May 2009 13:27:13 -0500 Subject: ZCM 10.2? In-Reply-To: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> Message-ID: <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> Last I heard was either today or tomorrow... That was on Friday. Patrick A. Hasenjager Network Administrator Kansas City University of Medicine and Biosciences phone 816.283.2478 fax 816.283.0692 email phasenjager at kcumb.edu >>> On 5/26/2009 at 1:13 PM, "Patrick Schmidt" wrote: Does anyone have any insight on SP2 for ZCM will be released? -------------------------------------------------- Patrick Schmidt Millard Public Schools Network Support Specialist (402)715-6278 peschmid at mpsomaha.org --------------------------------------------------- _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From bbrush at gmail.com Wed May 27 16:56:50 2009 From: bbrush at gmail.com (Bill Brush) Date: Wed, 27 May 2009 10:56:50 -0500 Subject: Travelogue - Adventures in OES - cluster question Message-ID: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> Ok, gents, as we covered a while ago, I am transitioning from my venerable, but aging Netware cluster to a new and improved spiffy shiny, super-cool OES-L cluster. "The Plan" calls for me to add Linux nodes to the existing cluster, then drop out the Netware nodes as they become unneeded. Before I charge with reckless abandon into my production environment I am building a VMware test case. So far I've been able to get a tree up and running and I can connect to NCP shares on Linux without a problem. Now when I go to cluster things I find myself chasing my tail, and I know some of it is self-inflicted. I have a Netware box, and I have a Linux (OES) box. When I create a cluster with the Linux box I can't add the Netware box in later. It says there isn't a cluster there to join even though the Linux box is happily reporting that there is. Ok, they probably never intended you to add a Netware box to a Linux cluster, so it probably doesn't work. So I nuked that cluster. When I try to create one on the Netware box it won't create one because there's only one Netware node (using NWdeploy.) I know I saw a cool solution on how to do this, but now I can't find it. So other than creating another Netware box, does anyone know how I can get this cluster created? Thanks! Bill From hooeld at bay.k12.fl.us Wed May 27 17:06:18 2009 From: hooeld at bay.k12.fl.us (Leslie Hooe) Date: Wed, 27 May 2009 11:06:18 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> Message-ID: <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> I did not think you could mix OS's in a cluster.. Leslie Hooe Telecommunications Manager Bay District Schools (850) 747-5295 >>> On 5/27/2009 at 10:56 AM, in message <167f4090905270856hc733e1cq218bd857794b1289 at mail.gmail.com>, Bill Brush wrote: Ok, gents, as we covered a while ago, I am transitioning from my venerable, but aging Netware cluster to a new and improved spiffy shiny, super-cool OES-L cluster. "The Plan" calls for me to add Linux nodes to the existing cluster, then drop out the Netware nodes as they become unneeded. Before I charge with reckless abandon into my production environment I am building a VMware test case. So far I've been able to get a tree up and running and I can connect to NCP shares on Linux without a problem. Now when I go to cluster things I find myself chasing my tail, and I know some of it is self-inflicted. I have a Netware box, and I have a Linux (OES) box. When I create a cluster with the Linux box I can't add the Netware box in later. It says there isn't a cluster there to join even though the Linux box is happily reporting that there is. Ok, they probably never intended you to add a Netware box to a Linux cluster, so it probably doesn't work. So I nuked that cluster. When I try to create one on the Netware box it won't create one because there's only one Netware node (using NWdeploy.) I know I saw a cool solution on how to do this, but now I can't find it. So other than creating another Netware box, does anyone know how I can get this cluster created? Thanks! Bill _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Under Florida law, e-mail addresses are public records. If you do not want your e-mail address released in response to a public-records request, do not send electronic mail to this entity. Instead, contact this office by phone or in writing. From Rpd at co.mason.wa.us Wed May 27 17:11:11 2009 From: Rpd at co.mason.wa.us (Bob Deans) Date: Wed, 27 May 2009 09:11:11 -0700 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> Message-ID: <4A1D03AD.6CD7.0020.1@co.mason.wa.us> You can mix cluster os's. http://www.novell.com/communities/node/5707/oes-netware-and-linux-mixed-os-clusters Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> "Leslie Hooe" 5/27/2009 9:06 AM >>> From geoffreycarman at gmail.com Wed May 27 17:13:34 2009 From: geoffreycarman at gmail.com (Geoffrey Carman) Date: Wed, 27 May 2009 12:13:34 -0400 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> Message-ID: <993788ac0905270913x56db7fc5va39fe71d3a257f61@mail.gmail.com> If I recall correctly it is the other way around. You can add Linux nodes to a Netware cluster, but not a Netware node to a Linux cluster, and there is some restriction on the disk space that I forget the details of. On Wed, May 27, 2009 at 12:06 PM, Leslie Hooe wrote: > I did not think you could mix OS's in a cluster.. > > Leslie Hooe > Telecommunications Manager > Bay District Schools > (850) 747-5295 > >>>> On 5/27/2009 at 10:56 AM, in message <167f4090905270856hc733e1cq218bd857794b1289 at mail.gmail.com>, Bill Brush wrote: > > Ok, gents, as we covered a while ago, I am transitioning from my venerable, > but aging Netware cluster to a new and improved spiffy shiny, super-cool > OES-L cluster. > > "The Plan" calls for me to add Linux nodes to the existing cluster, then > drop out the Netware nodes as they become unneeded. ?Before I charge with > reckless abandon into my production environment I am building a VMware test > case. ?So far I've been able to get a tree up and running and I can connect > to NCP shares on Linux without a problem. > > Now when I go to cluster things I find myself chasing my tail, and I know > some of it is self-inflicted. > > I have a Netware box, and I have a Linux (OES) box. ?When I create a cluster > with the Linux box I can't add the Netware box in later. ?It says there > isn't a cluster there to join even though the Linux box is happily reporting > that there is. ?Ok, they probably never intended you to add a Netware box to > a Linux cluster, so it probably doesn't work. ?So I nuked that cluster. > > When I try to create one on the Netware box it won't create one because > there's only one Netware node (using NWdeploy.) ?I know I saw a cool > solution on how to do this, but now I can't find it. > > So other than creating another Netware box, does anyone know how I can get > this cluster created? > > Thanks! > Bill > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > > > > > The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Under Florida law, e-mail addresses are public records. If you do not want your e-mail address released in response to a public-records request, do not send electronic mail to this entity. Instead, contact this office by phone or in writing. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- Geoffrey Carman geoffreycarman at gmail.com From hooeld at bay.k12.fl.us Wed May 27 17:14:29 2009 From: hooeld at bay.k12.fl.us (Leslie Hooe) Date: Wed, 27 May 2009 11:14:29 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <4A1D03AD.6CD7.0020.1@co.mason.wa.us> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> <4A1D03AD.6CD7.0020.1@co.mason.wa.us> Message-ID: <4A1D2095020000010021B0EB@BNGW.bay.k12.fl.us> I did not know that, I can see where that would be handy during some moves. I'll check it out. Thanks for the info. Les Leslie Hooe Telecommunications Manager Bay District Schools (850) 747-5295 >>> On 5/27/2009 at 11:11 AM, in message <4A1D03AD.6CD7.0020.1 at co.mason.wa.us>, "Bob Deans" wrote: You can mix cluster os's. http://www.novell.com/communities/node/5707/oes-netware-and-linux-mixed-os-clusters Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> "Leslie Hooe" 5/27/2009 9:06 AM >>> _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Under Florida law, e-mail addresses are public records. If you do not want your e-mail address released in response to a public-records request, do not send electronic mail to this entity. Instead, contact this office by phone or in writing. From clarridge at alan.ameslab.gov Wed May 27 17:22:19 2009 From: clarridge at alan.ameslab.gov (Mark Clarridge) Date: Wed, 27 May 2009 11:22:19 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> Message-ID: <4A1D226B0200001800039B55@alan.ameslab.gov> Is it possible to cluster enable a nss volume that I currently have on a stand alone Netware box and then install clustering and add a second node that's OES LINUX? This would be the least disruptive process I have in mind to migrate my legacy system. Great spirits have always encountered violent opposition from mediocre minds......... Einstein Mark Clarridge clarridge at ameslab.gov Ames Laboratory US DOE Iowa State University Ames IA 50248 >>> From: Bill Brush To: Novell LAN Interest Group Date: 5/27/2009 10:57 AM Subject: Travelogue - Adventures in OES - cluster question Ok, gents, as we covered a while ago, I am transitioning from my venerable, but aging Netware cluster to a new and improved spiffy shiny, super-cool OES-L cluster. "The Plan" calls for me to add Linux nodes to the existing cluster, then drop out the Netware nodes as they become unneeded. Before I charge with reckless abandon into my production environment I am building a VMware test case. So far I've been able to get a tree up and running and I can connect to NCP shares on Linux without a problem. Now when I go to cluster things I find myself chasing my tail, and I know some of it is self-inflicted. I have a Netware box, and I have a Linux (OES) box. When I create a cluster with the Linux box I can't add the Netware box in later. It says there isn't a cluster there to join even though the Linux box is happily reporting that there is. Ok, they probably never intended you to add a Netware box to a Linux cluster, so it probably doesn't work. So I nuked that cluster. When I try to create one on the Netware box it won't create one because there's only one Netware node (using NWdeploy.) I know I saw a cool solution on how to do this, but now I can't find it. So other than creating another Netware box, does anyone know how I can get this cluster created? Thanks! Bill _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From bbrush at gmail.com Wed May 27 17:25:45 2009 From: bbrush at gmail.com (Bill Brush) Date: Wed, 27 May 2009 11:25:45 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <4A1D226B0200001800039B55@alan.ameslab.gov> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D226B0200001800039B55@alan.ameslab.gov> Message-ID: <167f4090905270925m6a4b01ddnce4175452e99996e@mail.gmail.com> On Wed, May 27, 2009 at 11:22 AM, Mark Clarridge wrote: > > Is it possible to cluster enable a nss volume that I currently have on a > stand alone Netware box and then install clustering and add a second node > that's OES LINUX? This would be the least disruptive process I have in mind > to migrate my legacy system. > > > > This is what I'm trying to do. However! I have a shared disk which presumably your stand-alone Netware box would not have (being standalone). The problem I'm running into is how to create a mixed Netware/Linux cluster from scratch. To cluster a volume it has to be on some sort of sharable storage. Bill From clarridge at alan.ameslab.gov Wed May 27 17:32:15 2009 From: clarridge at alan.ameslab.gov (Mark Clarridge) Date: Wed, 27 May 2009 11:32:15 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <167f4090905270925m6a4b01ddnce4175452e99996e@mail.gmail.com> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D226B0200001800039B55@alan.ameslab.gov> <167f4090905270925m6a4b01ddnce4175452e99996e@mail.gmail.com> Message-ID: <4A1D24BF0200001800039B5A@alan.ameslab.gov> We do have a shared disk system in the form of an IBM fiber channel SAN. I have added the necessary hba card to the new server. I have not added or changed anything to my legacy Netware server. So I install NCS services on the NetWare server and then cluster enable the volume? Great spirits have always encountered violent opposition from mediocre minds......... Einstein Mark Clarridge clarridge at ameslab.gov Ames Laboratory US DOE Iowa State University Ames IA 50248 >>> From: Bill Brush To: Novell LAN Interest Group Date: 5/27/2009 11:26 AM Subject: Re: Travelogue - Adventures in OES - cluster question On Wed, May 27, 2009 at 11:22 AM, Mark Clarridge wrote: > > Is it possible to cluster enable a nss volume that I currently have on a > stand alone Netware box and then install clustering and add a second node > that's OES LINUX? This would be the least disruptive process I have in mind > to migrate my legacy system. > > > > This is what I'm trying to do. However! I have a shared disk which presumably your stand-alone Netware box would not have (being standalone). The problem I'm running into is how to create a mixed Netware/Linux cluster from scratch. To cluster a volume it has to be on some sort of sharable storage. Bill _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From Rpd at co.mason.wa.us Wed May 27 17:44:01 2009 From: Rpd at co.mason.wa.us (Bob Deans) Date: Wed, 27 May 2009 09:44:01 -0700 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <993788ac0905270913x56db7fc5va39fe71d3a257f61@mail.gmail.com> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> <993788ac0905270913x56db7fc5va39fe71d3a257f61@mail.gmail.com> Message-ID: <4A1D0B5F.6CD7.0020.1@co.mason.wa.us> http://www.novell.com/coolsolutions/feature/17240.html http://scottf.wordpress.com/2006/03/21/migrating-a-netware-cluster-to-an-open-enterprise-server-linux-cluster/ Converting Netware to Linux cluster instructions: http://www.novell.com/documentation/oes2/clus_admin_lx/index.html?page=/documentation/oes2/clus_admin_lx/data/ncscomparelxnw.html Robert Deans MCSE, CNE, CCENT, SECURITY+, Server+ LINUX+, NETWORK+, A+ IS Manager Mason County Washington 360-427-5503 >>> Geoffrey Carman 5/27/2009 9:13 AM >>> If I recall correctly it is the other way around. You can add Linux nodes to a Netware cluster, but not a Netware node to a Linux cluster, and there is some restriction on the disk space that I forget the details of. On Wed, May 27, 2009 at 12:06 PM, Leslie Hooe wrote: > I did not think you could mix OS's in a cluster.. > > Leslie Hooe > Telecommunications Manager > Bay District Schools > (850) 747-5295 > >>>> On 5/27/2009 at 10:56 AM, in message <167f4090905270856hc733e1cq218bd857794b1289 at mail.gmail.com>, Bill Brush wrote: > > Ok, gents, as we covered a while ago, I am transitioning from my venerable, > but aging Netware cluster to a new and improved spiffy shiny, super-cool > OES-L cluster. > > "The Plan" calls for me to add Linux nodes to the existing cluster, then > drop out the Netware nodes as they become unneeded. Before I charge with > reckless abandon into my production environment I am building a VMware test > case. So far I've been able to get a tree up and running and I can connect > to NCP shares on Linux without a problem. > > Now when I go to cluster things I find myself chasing my tail, and I know > some of it is self-inflicted. > > I have a Netware box, and I have a Linux (OES) box. When I create a cluster > with the Linux box I can't add the Netware box in later. It says there > isn't a cluster there to join even though the Linux box is happily reporting > that there is. Ok, they probably never intended you to add a Netware box to > a Linux cluster, so it probably doesn't work. So I nuked that cluster. > > When I try to create one on the Netware box it won't create one because > there's only one Netware node (using NWdeploy.) I know I saw a cool > solution on how to do this, but now I can't find it. > > So other than creating another Netware box, does anyone know how I can get > this cluster created? > > Thanks! > Bill > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > > > > > > The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Under Florida law, e-mail addresses are public records. If you do not want your e-mail address released in response to a public-records request, do not send electronic mail to this entity. Instead, contact this office by phone or in writing. > > > _______________________________________________ > Novell mailing list > Novell at netlab1.oucs.ox.ac.uk > http://netlab1.usu.edu/mailman/listinfo/novell > -- Geoffrey Carman geoffreycarman at gmail.com _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From MGlenn at cco.state.oh.us Wed May 27 21:10:11 2009 From: MGlenn at cco.state.oh.us (Michael Glenn) Date: Wed, 27 May 2009 16:10:11 -0400 Subject: migfiles vs shadowfs In-Reply-To: <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> Message-ID: <4A1D65E3.33FD.002C.1@cco.state.oh.us> Hey; if I'm using migfiles to migrate a bunch of NSS directory structures from Netware to an OES2 SP1 box running DST, should I set the target path to the master volume object in the /media/nss/ path, or should I target the object located in /media/shadowfs/? Thanks. From bbrush at gmail.com Wed May 27 21:25:17 2009 From: bbrush at gmail.com (Bill Brush) Date: Wed, 27 May 2009 15:25:17 -0500 Subject: Travelogue - Adventures in OES - cluster question In-Reply-To: <4A1D0B5F.6CD7.0020.1@co.mason.wa.us> References: <167f4090905270856hc733e1cq218bd857794b1289@mail.gmail.com> <4A1D1EAA020000010021B0DE@BNGW.bay.k12.fl.us> <993788ac0905270913x56db7fc5va39fe71d3a257f61@mail.gmail.com> <4A1D0B5F.6CD7.0020.1@co.mason.wa.us> Message-ID: <167f4090905271325x37ec50fcn5746cf904783587f@mail.gmail.com> Update: In the interest of getting the boat launched I just created another Netware box in VMware and created a 2 node Netware cluster and added the linux node into the the mix. Once I verify that my test resource migrates nicely I'll remove the superfluous Netware node. Bill From MGlenn at cco.state.oh.us Wed May 27 21:33:00 2009 From: MGlenn at cco.state.oh.us (Michael Glenn) Date: Wed, 27 May 2009 16:33:00 -0400 Subject: ZCM 10.2? In-Reply-To: <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> Message-ID: <4A1D6B3C.33FD.002C.1@cco.state.oh.us> ZCM_10.2.0.iso is now listed on my Customer Support page. >>> "Patrick Hasenjager" 05/26/2009 14:27 >>> Last I heard was either today or tomorrow... That was on Friday. Patrick A. Hasenjager Network Administrator Kansas City University of Medicine and Biosciences phone 816.283.2478 fax 816.283.0692 email phasenjager at kcumb.edu >>> On 5/26/2009 at 1:13 PM, "Patrick Schmidt" wrote: Does anyone have any insight on SP2 for ZCM will be released? -------------------------------------------------- Patrick Schmidt Millard Public Schools Network Support Specialist (402)715-6278 peschmid at mpsomaha.org --------------------------------------------------- _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joe.doupnik at oucs.ox.ac.uk Thu May 28 09:48:50 2009 From: joe.doupnik at oucs.ox.ac.uk (Joe Doupnik) Date: Thu, 28 May 2009 09:48:50 +0100 Subject: migfiles vs shadowfs In-Reply-To: <4A1D65E3.33FD.002C.1@cco.state.oh.us> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> <4A1BEE3102000005001ABC3A@smtp.kcumb.edu> <4A1D65E3.33FD.002C.1@cco.state.oh.us> Message-ID: <4A1E4FF2.60907@oucs.ox.ac.uk> Michael Glenn wrote: > Hey; if I'm using migfiles to migrate a bunch of NSS directory structures from Netware to an OES2 SP1 box running DST, should I set the target path to the master volume object in the /media/nss/ path, or should I target the object located in /media/shadowfs/? > ------------- Think about how shadowing works: the users see the merged view, not where files reside on physical storage. Modifying a file brings it into the foreground. Shadowfs is a different animal, a merged view at the VFS level for systems use. I suggest you leave that alone unless you have a very specific need. Also, it is not a heavy duty interface (uses FUSE). Thus migrate files to the appropriate fore/background areas as you think best. Joe D. From glenn.collrin at marwoodmail.com Thu May 28 20:54:14 2009 From: glenn.collrin at marwoodmail.com (Glenn Collrin) Date: Thu, 28 May 2009 16:54:14 -0300 Subject: migration of Netware 5.1 sp7 datafile to NTFS In-Reply-To: <4A1E4FF2.60907@oucs.ox.ac.uk> References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> <4A1BEE3102000005001ABC3A@smtp.kcumb.edu><4A1D65E3.33FD.002C.1@cco.state.oh.us> <4A1E4FF2.60907@oucs.ox.ac.uk> Message-ID: I am currently evaluating shutting down one of our Netware 5.1 sp8 Novell Servers that is acting only as a file server. Anyone have any pointers, white paper, etc on file migrations to NTFS which minimizes the user disruption. I am in process of testing feasible solutions, and currently decided to restore from tape based on number of files and directories. I had to first ensure that backup system backed up the Novell Files in uncompressed format. I want to ensure right people have access to right files, currently use group and file and folder assignments in Novell, Sounding like a manual process for creating groups and assignments in "AD", groups may be manageable but how accommodate all the individual files and folder assignments. This part sounds tricky I do not have much flexibility to changing drive letters and paths since a lot of applications have these embedded. Anyone want to tell me how they approached their success stories, and might as well hear the horror stories so I know what to avoid. Glenn C is wishing he could stay with Novell. From James.Taylor at eastcobbgroup.com Thu May 28 21:26:51 2009 From: James.Taylor at eastcobbgroup.com (James Taylor) Date: Thu, 28 May 2009 16:26:51 -0400 Subject: migration of Netware 5.1 sp7 datafile to NTFS In-Reply-To: References: <4A1BEAFB020000E400072F72@groupwise.mpsomaha.org> <4A1BEE3102000005001ABC3A@smtp.kcumb.edu><4A1D65E3.33FD.002C.1@cco.state.oh.us> <4A1E4FF2.60907@oucs.ox.ac.uk> Message-ID: <4A1EBB4B020000750003405B@inet.eastcobbgroup.com> It will be ugly, based on the experiences that a number of my clients have had. Apparently you can ease some of the pai using the Quest migration tool, but it appears to be pricey and limited. I'll let others add details. -jt James Taylor The East Cobb Group, Inc. 678-697-9420 james.taylor at eastcobbgroup.com http://www.eastcobbgroup.com >>> "Glenn Collrin" 5/28/2009 03:54 PM >>> I am currently evaluating shutting down one of our Netware 5.1 sp8 Novell Servers that is acting only as a file server. Anyone have any pointers, white paper, etc on file migrations to NTFS which minimizes the user disruption. I am in process of testing feasible solutions, and currently decided to restore from tape based on number of files and directories. I had to first ensure that backup system backed up the Novell Files in uncompressed format. I want to ensure right people have access to right files, currently use group and file and folder assignments in Novell, Sounding like a manual process for creating groups and assignments in "AD", groups may be manageable but how accommodate all the individual files and folder assignments. This part sounds tricky I do not have much flexibility to changing drive letters and paths since a lot of applications have these embedded. Anyone want to tell me how they approached their success stories, and might as well hear the horror stories so I know what to avoid. Glenn C is wishing he could stay with Novell. _______________________________________________ Novell mailing list Novell at netlab1.oucs.ox.ac.uk http://netlab1.usu.edu/mailman/listinfo/novell From joea at j4computers.com Fri May 29 16:32:46 2009 From: joea at j4computers.com (joea at j4computers.com) Date: Fri, 29 May 2009 11:32:46 -0400 Subject: OES1 linux, 4GB, "free" shows 3GB Message-ID: <4A1FF11F020000850005FF8F@FS-LIN-OES> OES1, Linux. HP box has 4 GB installed, "free -mg" shows 3 GB "total". ? joe a. From joe.doupnik at oucs.ox.ac.uk Fri May 29 17:16:43 2009 From: joe.doupnik at oucs.ox.ac.uk (Joe Doupnik) Date: Fri, 29 May 2009 17:16:43 +0100 Subject: OES1 linux, 4GB, "free" shows 3GB In-Reply-To: <4A1FF11F020000850005FF8F@FS-LIN-OES> References: <4A1FF11F020000850005FF8F@FS-LIN-OES> Message-ID: <4A200A6B.5000703@oucs.ox.ac.uk> joea at j4computers.com wrote: > OES1, Linux. HP box has 4 GB installed, "free -mg" shows 3 GB "total". > > ? -------- Util "free" can show all kinds of numbers but we are not certain how it arrived at them. The Gigabyte variety is especially prone to being rounded and such. Try just plain free and in keeping with open source thinking we be careful about the meaning of free (it does not mean total). Joe D.