Comment 3 for bug 1580745

glance-simplestreams-sync charm handles a bunch of "expected" exceptions, since they are thrown when all of the dependencies are not fully set up yet: it is designed to retry in such cases.

The core question here is if this was something that would have resolved itself with time, when it would have made sense to retry, or not. Do you perhaps have rest of the logs for this cloud deployment? Can you check the state of ceph-radosgw (unit state, logs in /var/log/ceph/radosgw.log, is radosgw service even running)? If you've still got the cloud up, a manual run after the issue has been resolved should then work:

  JUJU_HOME=/var/lib/landscape/juju-homes/1 juju ssh glance-simplestreams-sync/0 sudo /etc/cron.daily/glance-simplestreams-sync

I was going to make the script ignore and retry on all errors (since we now retry on a number of them: Chad just fixed another one), but if this one in particular will never lead to a success without manual intervention, there is no sense in retrying, and the problem is somewhere else.