In the first quarter of this year we started utilizing commvault backups for xenserver. They take a Xenserver snapshot and back up the device. Since then we periodically lose a guest and sometimes even a host due to failures during the snapshot process. I found a discussion about this on the Citrix forums where a recommendation for 30% LUN utilization is made. It sounds like you need the free space as Citrix sometimes utilizes double the space of a drive while coalescing a snapshot. This is ... insane.
So, a couple of questions:
1. Has anyone else run into this?
2. Does anyone know if, in a thin provisioned environment, the space is returned to an aggregate after coalescing is complete?
3. Has anyone seen documentation around this?
This forum chatter is difficult for me to test as I do not have the aggregate space to CYA if Xenserver decides to actually use all the "free" space. Anyone with experience surround Xenserver snapshot coalescence and free space requirements please help!
I can't comment on the outages as I don't know the Commvault product very well but it sounds like a problem quiecing the VMs since the NetApp snapshot will not interrupt the VMs at all.
The space issue I can comment on. This is a bit of an error in the settings that Citrix uses with it StorageLink product. They don't correctly set the fractional reserve on cloned LUNs which means on the first snapshot it consumes full space. I would recommend creating the LUNs using the new VSC for Xenserver instead (if you have NFS) or manually creating the LUNs if you are using Block with the correct thin provisioning settings to prevent this space grab,