Hi,
This weekend we've got an issue with a disk that did not get extend
anymore because of the following reason:
2022-12-17 12:12:27,852+0100 INFO (libvirt/events) [virt.vm]
(vmId='a2926e08-d19b-4e3a-98fa-cc1ee241a037') abnormal vm stop device
ua-6ccc3ee1-02e5-4fad-b1b7-9f2d6c187416 error enospc (vm:5185)
2022-12-17 12:12:27,852+0100 INFO (libvirt/events) [virt.vm]
(vmId='a2926e08-d19b-4e3a-98fa-cc1ee241a037') CPU stopped: onIOError (vm:6066)
2022-12-17 12:12:27,852+0100 INFO (libvirt/events) [virt.vm]
(vmId='a2926e08-d19b-4e3a-98fa-cc1ee241a037') drive 'sda' needs extension
(storage:963)
2022-12-17 12:12:27,854+0100 INFO (libvirt/events) [virt.vm]
(vmId='a2926e08-d19b-4e3a-98fa-cc1ee241a037') CPU stopped: onSuspend (vm:6066)
2022-12-17 12:12:27,931+0100 INFO (periodic/3) [virt.vm]
(vmId='a2926e08-d19b-4e3a-98fa-cc1ee241a037') Drive sda extended to maximum size,
disabling monitoring (thinp:357)
Now I looked into the VDSM code, and noticed that some years ago there
was a commit
https://github.com/oVirt/vdsm/commit/93e955af9e4989711403e81835bb469b2e9c...
Where the estimate of 1.1x the disk size was used as max size.
But is this 1.1x still valid? As for example since some years there are
dirty bitmaps etc which all take space in the qcow2 image.
As far as I see there is no way to really calculate the exact max size
(as it might depend on external factors), but it might be useful to
extend the 1.1x to 1.2x for example?
Or maybe the limit is not needed anymore at all?
Thanks
Jean-Louis