Unsolved
This post is more than 5 years old
9 Posts
0
468
July 4th, 2011 09:00
Migration of composite groups
I'm trying to migrate some devices into new composite groups.
For various reasons, we're wanting to start using -rdf_consistency, where we weren't before.
And ideally, in the process, start merging a few CGs, so we've got less.
The problem is, this is currently part of a global cluster, so what I could really do with doing, is creating _new_ CGs, making them 'ready' handing them over, and then tidying up the old ones after.
I'm having a bit of difficulty though - symcg copy/copyall appears to have the functionality I require.
However, every time I've tried using it (SYMAPI version 7.1-7.3) I get errors along a theme of "The device is already a member of a CG"
I've set the SYMAPI options file to include:
SYMAPI_ALLOW_DEV_IN_MULT_GRPS = ENABLE
But still no joy.
Any one able to shed any light on what I might be missing?
I've not got anything complicated in terms of CG contents - just devices (R1s or 2s depending on which site we're talking).
Although just to keep things interesting, global name serveice and group mirroring are part of the mixture.
CGs are for RDF.
symcg move and moveall seems to work, but I can't seem to get copy/copyall doing the same thing.
Please tell me there's something trivial I'm missing?
EdRolison
9 Posts
0
July 4th, 2011 09:00
And the bonus question - I'm enabling RDF consistency, in an environment where there's only two arrays (R1 and R2).
The bonus question is
a) is this utterly pointless anyway. (Host LUNs will be spanning disks, but we are using our CGs 'like' DGs - we went with CGs because we originally had different 'sets' of disks we wanted to do this with - a moot point now we're trying multiple DG occupancy)
b) If it's pointless, is there any way to tell the veritas clustering software to _not_ try to do a 'symcg enable/disable' when doing a failover on the CGs? (Other than migrating to device groups)
EdRolison
9 Posts
0
July 4th, 2011 09:00
Bit of an update.
Seems like this behaviour only exists when I'm doing this with '-rdf_consistent'.
That's really quite irritating. as the obvious workaround won't do what I'm trying to sort - specifically, to migrate from composite groups to consistency groups.