In theory if you're using snapshots you're fine.
If your backups are really taking 12 hours I'd start looking at
adding spool disks to speed them up as that's (obviously) a long time,
at least for nightlies, weeklies might be ok.
Hi, I wonder what will happen when the VMs are restarted while backup is running ?
Our backup runs almost 12 hrs at night. And at that time, we are also going to apply the patches and reboot the VMs. I am not sure what will happen if the backup is running and we reboot the VM.
We are using VCB for our snapshots.
Thanks
In theory if you're using snapshots you're fine.
If your backups are really taking 12 hours I'd start looking at
adding spool disks to speed them up as that's (obviously) a long time,
at least for nightlies, weeklies might be ok.
Starwind Software Developer
I agree with the last post. It will be fine, just some extra disk load.
The Snaps re-direct writes to another VMDK to the original disk is preserved at the point of snapshot. The concern is thou, that the longer your backups take, the bigger the snapshot delta will be. Depending on the delta rate the snapshots could become large and take a while to fold in.
If your patching/updating machines during the snapshot period the delta's will grow even faster. Make sure the datastores have plenty of free space and monitor the I/O load.
cheers and please flick points if that helped
Depends.
If you are backing up the VM store, you should be OK if you are
using some kind of snapshotting. Otherwise you are backing up
inconsistent data anyways.
If you are backing the VMs up as if they were real computers, then
your backup will probably abort if the VM reboots while the backup is
happening.
StarWind Software R&D
We are using VCB for our snapshots.
If the backup start BEFORE the upgrade, then the snap will give a "consistent" backup.
But if backup start during the reboot maybe you can have some problem: for example if you have guest quiescence scripts they could failure.
Andre