Keeping Global Service Groups in Sync

[Brian_Harvey]

Under VCS 5.0 with GCO and global service groups (SG in production cluster linked to a SG in a remote failover cluster), I have the requirement to come up with a way to keep these global SG's in sync. If somebody adds a new resource to the production SG, then the same change needs to be made in the "failvoer" SG to ensure a valid recovery during failover.

Has anyone come up with slick solutions for this? Either with a script which can copy a SG from one cluster to another or a script with can "diff" the SGs between two clusters to at least identify that that SGs are out of sync?

- Brian

[Gene_Henriksen]

Part of that problem can be looked at from the FireDrill. If you add a tablespace to a database and do not replicate the new tablespace, the FireDrill will fail. In Unix, the FireDrill can be run via cron and so could be used nightly to test the ability of the application to come up. IF doing this in UNIX, make sure it is in the cron on every system in the cluster. The reason is that the FD script (fdsched) will only run on the lowest numbered running node in the cluster. If you have node 0 and 1, run fdsched on 1 and iit will silently exit when it discovers 0 is running. That prevents multiple copies of the FD from trying to come online.

As far as keeping the configs for the SGs in sync, I don't have a solution.