Juju fails to find/open /dev/vdb in ceph-osd units

[Camille Rodriguez]

When running the test test_blocked_when_non_pristine_disk_appears (in zaza/openstack/charm_tests/ceph/tests.py), it appears that ceph-osd deployed with the disco series does not detect the non-pristine disk.

The expected workload/status would be
'workload-status': 'blocked','workload-status-message': 'Non-pristine

But the ceph-osd actually goes into error with hook-failed "config-changed".

In ceph-osd/0:
2019-09-03 14:34:20 DEBUG config-changed partx: /dev/vdb: failed to read partition table
2019-09-03 14:34:20 INFO juju-log Can't get info for /dev/vdb: b''
2019-09-03 14:34:20 DEBUG config-changed Failed to find physical volume "/dev/vdb".
2019-09-03 14:34:21 DEBUG config-changed Failed to find physical volume "/dev/vdb".
2019-09-03 14:34:21 DEBUG config-changed Can't open /dev/vdb exclusively. Mounted filesystem?

See also sos-report attached to this bug.

This has been hit in OSCI for charm-ceph-mon and charm-ceph-osd. i.e https://review.opendev.org/#/c/678647/ https://review.opendev.org/#/c/678648/

https://openstack-ci-reports.ubuntu.com/artifacts/test_charm_pipeline_func_full/openstack/charm-ceph-mon/678647/1/3780/index.html

[Camille Rodriguez (camille.rodriguez)]

I recommend to un-duplicate these bugs, since here, the issue occurs during the non-pristine test, but the issue is with juju being unable to find /dev/vdb, while the bug #1840836 is actually a failure with non-pristine devices being detected.

[Alex Kavanagh (ajkavanagh)]

Setting to incomplete as this is probably a test environment issue, rather than a charm issue; there have been and continue to be changes in juju to resolve this; please reset to new if the problem re-occurs.

[Launchpad Janitor (janitor)]

[Expired for OpenStack ceph-osd charm because there has been no activity for 60 days.]

[Launchpad Janitor (janitor)]

[Expired for OpenStack ceph-mon charm because there has been no activity for 60 days.]