Comment 0 for bug 1584938

This was seen on a Landscape Autopilot deployment (using ceph/ceph) after the fix for lp:1583276 was committed:

root@juju-machine-0-lxc-1:/var/lib/landscape/juju-homes/1# juju ssh glance-simplestreams-sync/0 apt-cache policy python-simplestreams
Warning: Permanently added 'node-11.vmwarestack,10.245.202.81' (ECDSA) to the list of known hosts.
Warning: Permanently added '10.245.201.73' (ECDSA) to the list of known hosts.
python-simplestreams:
  Installed: 0.1.0~bzr434~trunk-0ubuntu1~ubuntu14.04.1
  Candidate: 0.1.0~bzr434~trunk-0ubuntu1~ubuntu14.04.1

The glance-simplestreams-sync.log [1] shows that the xenial image was not completely updated and got stuck in the queued state:

WARNING * 05-23 16:29:08 [PID:18247] * sstreams * Ignoring inactive image 37d8dfe7-f829-48e0-a66f-aff0cbe0a201 with status 'queued'

The log then shows that another xenial image was downloaded and ultimately the sync process switches to the daily cron.

The problem occurs with the user tries to juju bootstrap with trusty. It appears that the trusty simplestreams metadata is incomplete [2,3] and leads to a failed bootstrap [4]. Creating a trusty instance via horizon (no juju involved) works fine and a bootstrap with xenial works also.

Attached are a collection of logs including all of the pastebins linked here.

[1] - https://pastebin.canonical.com/157049/
[2] - https://pastebin.canonical.com/157058/
[3] - https://pastebin.canonical.com/157059/
[4] - https://pastebin.canonical.com/157060/