ODX support?

[Colin Benson]

I am seeing a situation on a few customer sites running HP storevirtual servers where checkpoints taken as part of the backup process are hanging the VM and causing a loss of network connectivity to the VM.

No other servers are affected during the checkpoint and the hosts etc remain responsive it appears isolated to the VM where the checkpoint is happening.

I note that ODX is still enabled on both Windows 2012 R2 and Windows 2016 hosts - should I disable this or has anyone else seen this issue?

Using DPM as the backup solution.

[oikjn]

Are these hyper-v or esx hosts?  You mention DPM so i'll guess hyper-v.  Are you using the application aware snapshot agent on the hosts, and are they up to date?  Are you using the MS DSM or the HP DSM?  Does the snapshot say that its app aware or does it not say anything.  This is most likely the issue, but I"m not even sure what you are talking about with ODX, but my guess is the above is really more likely the issue.

[Colin Benson]

HyperV hosts using HP DSM.

Snapshot agent hasn’t been installed so I guess this is my first step.

ODX is offline data transfer which is enabled by default on 2012 r2 & 2016 hosts, I know support for this was not there in left hand days so was wondering is it supported now or should this be disabled?

VSA units are 12.6 so should i update to 12.7, have been wary of doing that.

[oikjn]

yes I would suggest you install the application managed snapshot program and select that as the snapshot option in CMC.  

As for 12.6 upgrade to 12.7.  12.7 has been out for a bit so I don't think there are any major issues with it, but some people like to stay at least one revision behind because they see them as generally less risky.

Do you have single VLUNs on single VMs?  You mention that one VM crashes and it seems to imply its the only one attached to that LUN. CMC does pause communication on the LUN while the initiation of the snapshot is generated, so depending on the VSA performance capability, the load on the vsas and the number of snapshots happening, this delay might be longer than that VM allows and it times out.