Restore ignite on a (not active) cluster

[Hugo Wilhelmus]

I want to restore from a ignite tape my HU-UX 10.20 on a K370 Server, which was at that time part of a cluster with an other K370.
This cluster is not active anymore but my other K370 is still a production-machine.
Might i encounter problems when restoring the ignite-tape (which also contains cluster-commands) ? Or can i restore without any problems. All my disks (including root-disks) are part of a shared-cab.

[Mel Burslan · Accepted solution]

Hugo,

When you say you will restore from ignite tape, are you refering to the ignite tape made on this, same server or is it the ignite tape of the other k370 which is the production server right now ?

If the case is the first one, i.e., restoring from self ignite image, I do not see any problems arising right off the bat. Most probably, after the restore, it is a good idea to make the first boot into single user and disable cluster from starting by editing the /etc/rc.config.d/cmcluster and by setting the variable AUTOSTART_CMCLD to 0. Then boot normally to your default init level. While the cluster is not running, there is no danger of corrupting the (once)shared disks. Although, you need to exercise extreme caution, as most probably the volumes your current production box is seeing will also be visible to this server as available volumes unless you are using SAN disks and have some zoning preventing this mishap. So when you execute LVM commands, be very, very, but *very* (I can not emphasize it enough) careful.

If you are igniting this server from an ignite image of the currently in production K370 box, i.e., the latter case I mentioned in my first paragraph, make sure you give it a new ip address to preven an unexpected downtime for your production machine if it is still using the same ip address that it had while it was a cluster member.

Hope this helps.

[Juan M Leon]

Hugo,

I will add one more thing to Mel suggestion, if you inginite from a different server, then after change the ip Address I will double check the configuration files of each package under /etc/cmcluster/{packagename} then run cmcheckconf -v to ensure taht all config files are correct and at the end run cmapplyconf -v to push the binaries to the alternate node. I will do this just to be safe.

Hope it helps

[Mel Burslan]

Juan, I think, what hugo wants is not to reform the cluster and rejoin this node to it but just bring the server up with the ignite image taken at the times when this node was a cluster member but now he wants it to be standalone. In which case it is unnecessary to recompile the cluster binaries or do anything cluster related, maybe other than removal of cluster components.

[Juan M Leon]

I missunderstood the question. Yep I agree with you

Thanks

[Hugo Wilhelmus]

Yes as Mel already said, i just want to downgrade my UX 11 to the old 10.20 but i'm not planning to reconfigure my old cluster.

It's just that the ignite is from 2002 from the same K370 i now want to downgrade.
So it's a good idea to editing the /etc/rc.config.d/cmcluster and by setting the variable AUTOSTART_CMCLD to 0
to avoid any problems.

About my current shared volumes : i have an good overview of what is now shared with my other production K370.

By the way why is ignite refering to 3 disk which will be used. In my current config i have 4 disks for vg00 (mirrored !)