[Juju/MAAS 2.0] Juju should release the node when provisioning fails due to missing images.

[Andres Rodriguez]

I tried to deploy a charm in Trusty as a series, knowing that it will fail because there was no Trusty image in MAAS.

ubuntu@trusty-maas9:~/charms$ juju deploy ./trusty/virtme --series trusty

While I knew the deploy would fail, the error's that I got did not really reflect the issue.

For example, the machine in MAAS remained 'Allocated', and juju status shows:

ubuntu@trusty-maas9:~/charms$ juju status
[Services]
NAME STATUS EXPOSED CHARM
virtme unknown false local:trusty/virtme-11

[Units]
ID WORKLOAD-STATUS JUJU-STATUS VERSION MACHINE PORTS PUBLIC-ADDRESS MESSAGE
virtme/0 unknown allocating 0 Waiting for agent initialization to finish

[Machines]
ID STATE DNS INS-ID SERIES AZ
0 error pending trusty

I'd have expected that under [Unites] the JUJU-STATUS section would show an error instead of 'allocating'.

Under the [Machines] section, while 'state' shows 'error', the INS-ID show's pending, but the deployment has obviously failed.

At the time logs I see in MAAS:

==> /var/log/maas/maas.log <==
May 13 04:58:18 trusty-maas9 maas.api: [INFO] Request from user admin to acquire a machine with constraints <QueryDict: {'zone': ['default'], 'agent_name': ['3c54187a-3605-47c5-87f9-e2032fb6ad28']}>
May 13 04:58:18 trusty-maas9 maas.node: [INFO] monocyclic-florentina: Status transition from READY to ALLOCATED
May 13 04:58:18 trusty-maas9 maas.node: [INFO] monocyclic-florentina: allocated to user admin

I also feel that if juju failed, it should have released the machine and not left it 'Allocated'

[Cheryl Jennings (cherylj)]

Bug #1575283 covers the case where the cause of errors are hidden in the default tabular output.

I'm going to make this bug just for releasing the machine when the provisioning failed.

[Canonical Juju QA Bot (juju-qa-bot)]

This bug has not been updated in 5 years, so we're marking it Expired. If you believe this is incorrect, please update the status.