VM timeouts during snapshot commit

[rgs3]

Every time that a backup/replication job runs and completes my virtual machines are unresponsive for about 30 seconds during the snapshot commit. This does NOT happen when I manually snap and commit/delete the snap, but only when Veeam backup server does it. I've been told that this is due to the fact that we are using NFS for storage. Can this be avoided?

Does this happen with everyone with NFS storage? Most of my backups run during the night so this isn't that huge of an issue, however when the backups spill into the morning lots of complaints from users that exchange/file servers ect are unresponsive.

Thanks in advance,
Richard

[Gostev]

Richard - actually Veeam Backup uses the same API call to initiate snapshot removal as when you do it manually with VMware Infrastructure Client (RemoveSnapshot function) and cannot affect or change this process anyhow. Our tool only issues this single command, while everything else is handled by ESX server itself - VMware does not give vendors more control than that with snapshot removal .

Generally, for any troubleshooting involving manual snapshot removals, you need to ensure 2 things:
a) Create quiesced snapshot (which what Veeam Backup does, but this is not a default option in VMware Infrastructure Client).
b) Keep snapshot open for long enough time before deleting it, similar to time it takes to backup the VM - to make sure it grows large (commiting large amounts of data into VMDK is what may cause some VM slowness).

Overall, this issue with timeouts should not happen in normal conditions. I recommend investigating VM log file for stun/unstun cycles length on snapshot commit. They would show if the issue is actually related to snapshot commit process.