When Trying to clone the Fedora-32 TemplateVM, after two minutes, the operation is aborted with the message Error while cloning quebe, file pool cannot export dirty volumes. This happens independently from having AppVMs based on this template running or not.
It also seems that cloning is much slower in R4.1 than in R4.0, but this may be a personal impression.
For the legacy ‘file’ (not ‘file-reflink’) storage driver, it’s generally unsafe to clone or a back up a running VM. R4.1 has a new check to prevent this - which will also trigger if the VM wasn’t shutdown cleanly, so you may have to restart it and then shut it down again if that’s the reason.
You can install R4.1 on Btrfs or XFS and it will automatically create the /var/lib/qubes storage pool using the modern ‘file-reflink’ driver, which does not have these limitations.
The error persists, although the VM is not running.
I’ll check the environment later on.
Did you try starting the VM and shutting it down again? This should ensure that it’s “clean”, and not “dirty” from a previous crash.
Yes - I started the VM and shut it down without error. Still, it cannot be cloned.
I suppose that my pool configuration was incorrect see Qubes issue #6329. After correcting this, template cloning works. So this problem is solved.
This is 100% reproducible on Qubes 4.1 still. Templates cannot be cloned, even if they are fully powered off.
The file-reflink driver is useless to those of us who are using ZFS.
https://github.com/QubesOS/qubes-issues/issues/7325 is the issue. It looks like an altogether different problem not fixed by the original fix.