Backup fleecing: stick to same storage automatically?

[cluck]

Backup fleecing is a good thing.

But I'm a bit puzzled by the need to choose a fleecing storage in the backup job:



I wonder if I'm the only one missing the option to configure fleecing to stick to the same storage the source is in, whatever that is.

I am thinking of having two options:

• the option on the storage to "allow allocating fleecing storage on this storage automatically", and
• the option on the backup job to "automatically choose fleecing storage" and fall back to the fleecing storage configured in the backup job only as needed

 

[fiona]

Hi,
the reason this is not currently possible is that certain storages are pretty bad as fleecing storages, e.g. with LVM you need to allocate a disk of the same size as the original disk up front! Network storages are also recommended against.

It is recommended to use a thin-provisioned, fast local storage which supports discard. A node-wide default can be configured in /etc/vzdump.conf by adding a line like: fleecing: enabled=true,storage=local-lvm

 

[cluck]

Indeed I was considering the fact that some storage types are not as suitable as others. I was also thinking of not proposing a change which would surprisingly change behavior for other People.

That's why I am proposing to introduce both a setting on the storage and on the backup job, both defaulting to off.

To enable the "same storage" mechanism would be a deliberate choice to make both on the storage and on the backup job. If one of the two, or both are not enabled, or the option isn't at all available on some type of storage (thus assuming it is disabled) then it would work just as it does now. Only if both are enabled, it would use the same storage.

Just to describe where I come from: I have multiple tenants on various clusters and each have their dedicated slow, fast and archive storage to choose from. Users can only choose from their tenant storage. VMs are usually hooked up to multiple of them. All of this storage is fast enough to support snapshots/copy-on-write activity (i.e. similar access pattern as with backup fleecing). So I don't really have a default per node and no default per VM that fits well. Adding a local storage to every node that matches the availability/consistency/integrity criteria of the shared storage is just practically impossible.

 

[fiona]

So just that I understand correctly: you have a single backup job on a given node that covers multiple such VMs with disks on different tenant storages? And splitting the job (grouping by different fleecing storage) is also not feasible?

I do think such or a similar setting could make sense then. Feel free to open a feature request on the bug tracker (linking back to here) so we can better keep track of it: https://bugzilla.proxmox.com/