#SnapDrive Windows Cluster issue...

1 messages · Page 1 of 1 (latest)

stone jackal
#

I am trying to setup a test environment of a Windows Server 2012R2 cluster, which we need to do a rolling update to 2016...
We are using the good old SnapDrive 🙂 It is cluster aware... 🙂
We can setup the vserver connection, and establish an iscsi session via SnapDrive...
But as we try to create a cluster aware disk it fails at the last part where you have to map the iscsi connections for each host...

The iSCSI node name 'iqn.1991-05.com.microsoft:fs01' of an iSCSI session established through 'ROOT\ISCSIPRT\0000_0'-'Microsoft iSCSI Initiator' iSCSI HBA does not match the iSCSI node name 'iqn.1991-05.com.microsoft:fs01.bbaas.local' of the Microsoft iSCSI service. These names must match for correct Snapdrive operation.

As you can see the initiator name has changed, because we first installed snapdrive when the server was not a member of an AD... we then added the servers to the AD, and the initiators changed names...
For some reason it remembers the old name... even though it isn't listed in the iscsi initiator and in none of the .config files in Snapdrive...

Also the SnapDrive service is stopped as part of the LUN creation... any good suggestions? I have of cause also searched the registry which also doesn't find anyting...

coral acorn
#

the proper way would probably be to disconnect the LUNs and clean up the stale session information in the registry. I think there's a KB for that (although I can't find it right now). I remember we had to regularly clean hundreds of these stale sessions from the registry on our backup verification servers as they would remember each and every LUN they ever connected (which is a lot for verification servers)

stone jackal
#

Yeah this is just a test system, so it's pretty easy to clear the session from either snapdrive or the iscsi initiator... or does it require even more to remove it? Tomorrow I will try to re-install snapdrive just to see if it fix it...

coral acorn
#

we had a cleanup script somewhere.. I can try and take a look next week if I can find it. I think it was just removing stale sessions but I don't remember under what key

stone jackal
#

...OK tried to remove snapdrive and all traces... then reinstalled it again... same issue, but this time I don't get the iSCSI node name issue... but the snapdrive service on "the other node" dies as I try to map a clustered drive... I managed to get it to complete by restarting the service... of cause I get the "standard" Event 1000 which just doesn't say anything... I could look in the snapdrive logs etc... but right now I got a cluster running with a few mapped LUNs... and I just hope the snapdrive behaves when I add a 2016 node to the cluster... not sure if it's possible to alter the igroup from snapdrive to include the new nodes addess to the LUNs.. or if I have to do it manually... this is all a test so that the "real" rolling update becomes as smooth as possible... Fun fact is that we don't have snapdrive/snapmanager licenses on the existing cluster, so we are using their old snapdrive licensekeys... but SnapCenter would have been nice to use instead, yet... no licenses.. so I guess we will have to do this manually in the future... (yes this is one of those customers who choose to reject ONTAP One, to save a few bucks...) 😉