Re: [zones-discuss] VCS failover of non-global zones between systems.
So basically today it sounds like we'll still need to build 2 seperate zones on each physical server for failover purposes. Especially in patching situations where I see this is the only workaround, unless someone has devised a way to only have one zone built and failed over via a SAN volume on each physical node and patching would not be an issue. I can't seem to find anything in the VCS 5.0 documentation that includes the attach/detach commands only some discussions that say this may be included some time in the near future. I'm assuming this is the same with SUN cluster 3. This message posted from opensolaris.org ___ zones-discuss mailing list zones-discuss@opensolaris.org
Re: [zones-discuss] VCS failover of non-global zones between systems.
Peter Wilk wrote: This implies that both (all) systems in the cluster MUST be at identical patch levels. Yes, you want this. Does Sun support the migration of zones from one machine to another via this technique? Is there an official position? A note that Sun Cluster 3.1 08/05 offers what appears to be similar to what you describe - so seems to be a common way to go with current capabilities... http://docs.sun.com/app/docs/doc/819-2664/6n4uhp5gm?q=zone+patcha=view /jason Thanks Peter ___ zones-discuss mailing list zones-discuss@opensolaris.org ___ zones-discuss mailing list zones-discuss@opensolaris.org
Re: [zones-discuss] VCS failover of non-global zones between systems.
Hi, This procedure was changed in the release notes. Modifying /etc/zones/index or the zonname.xml is unsupported. Keep in mind, that the comments in these files tell you to keep your hand off. Kind regards Detlef Ulherr On Fri, May 12, 2006 at 10:23:42AM -0400, Jason Schroeder wrote: Peter Wilk wrote: This implies that both (all) systems in the cluster MUST be at identical patch levels. Yes, you want this. Does Sun support the migration of zones from one machine to another via this technique? Is there an official position? A note that Sun Cluster 3.1 08/05 offers what appears to be similar to what you describe - so seems to be a common way to go with current capabilities... http://docs.sun.com/app/docs/doc/819-2664/6n4uhp5gm?q=zone+patcha=view /jason Thanks Peter ___ zones-discuss mailing list zones-discuss@opensolaris.org ___ zones-discuss mailing list zones-discuss@opensolaris.org -- ~~ NOTICE: This email message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message. ~~ * Detlef Ulherr Project Engineer Tel: (++49 6103) 752-248 Client Solutions Fax: (++49 6103) 752-167 Sun Microsystems GmbH Amperestraße 6 mailto:[EMAIL PROTECTED] 63225 Langen http://www.sun.de/ * ___ zones-discuss mailing list zones-discuss@opensolaris.org
Re: [zones-discuss] VCS failover of non-global zones between systems.
Peter, Personal experience For now and until zone migration options (detach/attach) are available, we have created primary/standby zones (same names and configurations on all failover nodes) with one condition where zones are in the following states: primary=RUNNING and all standby=INSTALLED. VCS will use its ZONE agent to online/offline/clean/monitor zone's status and its application entry-points done with "hawizard" veritas VCS 4.1 tool. Once the zone migration options (detach/attach) are available to the public, same all same OS rules of engagement on cluster nodes (Patches, Packages, ..etc) would have to be stressed out and Zones would be created on their own Shared Storage (1 Volume/1 DiskGroup) for the zone content itself to migrate, so we could experience steps like: Shutdown Application within Zone umount all application NFS shared volumes related to zone's application (NAS case) Halt the Zone to INSTALLED state Detach The Zone (zoneadm -z zonename detach) umount all application volumes related to zone's application (SAN case) Deport All DiskGroups for both Applications and Zone. Reverse the steps on the failed-over node and Attach the Zone (zoneadm -z zonename attach), boot the zone and start the application. At least that is what we have done so far and yet to experiment with zone migration options (detach/attach) shortly. Hope this helps cheers Ihsan Peter Wilk wrote: IHAC that is asking the following question from customer I have a question regarding the VCS failover of non-global zones between systems. Veritas supports a "zone" agent that is used to failover zones between systems under cluster control. There are several documented (by veritas) restrictions on this: (1) You must create a zone.xml file that is unique to cluster member, as well as update the index.xml file for the zone. (2) The zone root must be on shared SAN disk (managed by VxVM) so that it can be visible to both machines. My question is regarding Sun support of this technique. Due to the restrictions of non-global zones for patching and package-adding, when you patch (or add a package to) a system when the failover zone is present, it'll get patched. If it isn't present, it won't. This implies that both (all) systems in the cluster MUST be at identical patch levels. Does Sun support the migration of zones from one machine to another via this technique? Is there an official position? Thanks Peter ___ zones-discuss mailing list zones-discuss@opensolaris.org -- My2Signature Ihsan Zaghmouth Sr. SAP Solution Architect SUN-SAP Business Applications Group (832) 859-2818 (Cell) (713) 784-2818 (Home) (713) 784-2818 (Fax) [EMAIL PROTECTED] ___ zones-discuss mailing list zones-discuss@opensolaris.org
