As outlined before the time from device_del to the DEVICE_DELETED is indeed "a bit longer" in Focal (from ~0.1s to ~6s), but other than that I couldn't find anything else yet.
We were wondering due to [1] in a related bug if it might be dependent to (over)load.
I was consuming Disk and CPU in a controlled fashion (stress-ng --cpu 8 --io 4 --hdd 4 1x in the container the load is in and 1x on the Host). On that I was running attach/detach loops but in all of a two hundred retries it was "->device_del <-DEVICE_DELETED <-DEVICE_DELETED ->blockdev-del ->blockdev-del".
James Page was doing tests (thanks) with real Openstack and Ceph but couldn't - so far -trigger the issue either. He will likely later update the bug as well, but is currently trying to ramp up concurrency to more likely hit the issue.
As outlined before the time from device_del to the DEVICE_DELETED is indeed "a bit longer" in Focal (from ~0.1s to ~6s), but other than that I couldn't find anything else yet.
We were wondering due to [1] in a related bug if it might be dependent to (over)load.
I was consuming Disk and CPU in a controlled fashion (stress-ng --cpu 8 --io 4 --hdd 4 1x in the container the load is in and 1x on the Host). On that I was running attach/detach loops but in all of a two hundred retries it was "->device_del <-DEVICE_DELETED <-DEVICE_DELETED ->blockdev-del ->blockdev-del".
James Page was doing tests (thanks) with real Openstack and Ceph but couldn't - so far -trigger the issue either. He will likely later update the bug as well, but is currently trying to ramp up concurrency to more likely hit the issue.
https:/ /bugs.launchpad .net/nova/ +bug/1882521/ comments/ 8