There should be a way to prevent all checks from running and stop at the first failed check.
I suggest this because the check may take too long and the user will prefer a quick response over a detailed one. At the same time, there are checks that are bonded together. For example, if the Ceph cluster is not healthy due to a lack of ceph-osd units, then further checking whether the ceph-osd unit can be removed does not make much sense.
I think the best option would be to add the "--stop-on-failure" argument.
How does it work:
(juju-verify) ➜ juju-verify git:(bug/1917600-new) ✗ juju-verify reboot --units ceph-osd/0 ceph-osd/1
Result: FAIL
Reason: ceph-mon/0: Ceph cluster is healthy
The minimum number of replicas in 'ceph-osd' is 1 and it's not safe to restart/shutdown 2 units. 0 units are not active.
It's not safe to removed units ['ceph-osd/0', 'ceph-osd/1'] in the availability zone root=default. [free_units=1]
How it could work:
(juju-verify) ➜ juju-verify git:(bug/1917600-new) ✗ juju-verify reboot --units ceph-osd/0 ceph-osd/1 --stop-on-failure
Result: FAIL
Reason: ceph-mon/0: Ceph cluster is healthy
The minimum number of replicas in 'ceph-osd' is 1 and it's not safe to restart/shutdown 2 units. 0 units are not active.
esunar
commented
1 year ago
There should be a way to prevent all checks from running and stop at the first failed check.
I suggest this because the check may take too long and the user will prefer a quick response over a detailed one. At the same time, there are checks that are bonded together. For example, if the Ceph cluster is not healthy due to a lack of ceph-osd units, then further checking whether the ceph-osd unit can be removed does not make much sense.
I think the best option would be to add the "--stop-on-failure" argument.
How does it work:
How it could work:
Imported from Launchpad using lp2gh.
date created: 2021-03-25T21:19:26Z
owner: rgildein
assignee: rgildein
the launchpad url