The issues given by the bugged SRA plug-ins have finally ended (see here).
After a call openened with support, lot of logs exchanged, they've been able to reproduce the issue and development team reported:
"Development has confirmed an issue with thin provisioned volumes with AMGs. The only workaround is to avoid thin provisioned volumes for SRM until a release of SRA includes the fix for this issue. At this time, I do not have an ETA for the a release date of the SRA version that will include a fix.
For the time being, to get the customer into production, I recommend trying a non-thin provisioned volume within a Dynamic Disk Pool or creating a volume within a Volume group"
First I've created and paired in a new AMG a new thick volume but SRA failed the same nevertheless I put my old AMG (with thin vols) disabled. Only when I've completely removed the old AMGs I was able to complete the SRM device discovery and the creation of the protection group.
Well, mistery is solved but it remains a very big issue…a big issue because if you have also just one thin volume in a AMG the SRA fails to enumerate the paired devices!
I think that customer (and me too, I'm a “thin” evangelist) would use thin from arrays that claim they can do it, don’t you think so?
Well. If AMG in E-Series contain thin volumes the SRA plugin fails with the error I shared with you.
And, maybe this is worst, if you maintain also just one AMG with thin resources, SRA fails!
This means that in case you need just a couple of LUNs for SRM purposes all of your other DR resources, also the “crash consistancy ones”, must be THICK!
This is really crazy!
Anyway, my SRM/SRA worked only when I’ve created a new AMG with thick volumes and after I’ve removed all old thin mirrored resources and their AMGs.
It's really incredible to think that development team forgot the possibility to use thin volumes in AMG...using such array able to do thin! 😉
