Juju doesn't inform users when MAAS is out of nodes

[Jorge Castro]

To replicate, have a maas with X nodes. `juju add-unit -nX+1 service`, more than you really have.

MAAS knows that it is out of nodes and acts accordingly, but Juju just sits there and doesn't communicate with the user, the user then has to go to MAAS to find out what is really happening. Recommend something like:

    juju add-unit -n11 postgresql
    MAAS $clustername reports that only 10 nodes are available, provision more nodes via http://blah

Or something that hints that the user is out of resources.

[Dustin Kirkland  (kirkland)]

Well, technically, this information is available, in 'juju status':

agent-state-info: '(error: cannot run instances: gomaasapi: got error back from
  server: 409 Conflict (No matching node is available.))'
instance-id: pending

Now, at this point, I went and freed up a suitable resource in MAAS. However, Juju never actually checked back with MAAS to see if resources have become available. I had to manually run:

  juju retry-provisioning <machine-id>

This was awkward and unexpected. Juju should have checked back and done this automatically.

[Jorge Castro (jorge)]

Marco thinks MAAS 2.0 fixes this, we need to check when that integration work is complete.

[Anastasia (anastasia-macmood)]

There has been a lot of movement in the area in both Juju 2 and Maas 2.
If you are still seeing an issue, please file a bug in "juju" launchpad project as this is where we track bugs against Juju 2.