cancel
Showing results for 
Search instead for 
Did you mean: 

Rename disk group in VCS?

abelopez
Level 2

Hello,

I am in the process of converting a "DIY" HA solution to VCS. Due to limitations in the way the database is setup, I have to import the diskgroup and change its name, and when I deport the disk group, change it back. This was previously done with shell scripts calling vxdg with the -n option.

I can't seem to find a place for this option in VCS 5.1.

I need to be able to do this:

prod-rg, qa-rg

prod-rg imports db-prod diskgroup as db, deports diskgroup as db-prod

qa-rg imports db-qa diskgroup as db, deports diskgroup as db-qa

 

Any thoughts?

1 ACCEPTED SOLUTION

Accepted Solutions

mikebounds
Level 6
Partner Accredited

Firstly I would try and get round having to rename diskgroup.  If database is hardcoded to use raw devices then you might be able to use symlinks.  If this is not possible then you could rename diskgroup in a preonline and postoffline trigger script, but you may have problems with duplicate diskgroup names if a system crashes and so VCS does not run the postoffline, so I would try to use symlinks with uniquely named diskgroups, even if this means using in conjunction with a preonline if you have to move symlinks.

More information on your restrictions would be useful.  If you have 2 nodes and Prod contains same volumes as QA and QA shuts down when Prod fails over, then I think symlinks should work (along with an offline group dependency) without using preonline trigger

Mike

View solution in original post

2 REPLIES 2

mikebounds
Level 6
Partner Accredited

Firstly I would try and get round having to rename diskgroup.  If database is hardcoded to use raw devices then you might be able to use symlinks.  If this is not possible then you could rename diskgroup in a preonline and postoffline trigger script, but you may have problems with duplicate diskgroup names if a system crashes and so VCS does not run the postoffline, so I would try to use symlinks with uniquely named diskgroups, even if this means using in conjunction with a preonline if you have to move symlinks.

More information on your restrictions would be useful.  If you have 2 nodes and Prod contains same volumes as QA and QA shuts down when Prod fails over, then I think symlinks should work (along with an offline group dependency) without using preonline trigger

Mike

abelopez
Level 2

Yes, exactly, the databases are looking for raw devices, which the diskgroup name is part of the path. Symlinks may work, or even hard link. thanks