Comment 4 for bug 1673916

This appears to be because we're calling addCleanup() instead of addCallback() in the reportMDNSEntries() function in rackdservices/networks_monitoring_service.py.

I still didn't understand why this bug doesn't surface for everyone in the world (which is why I held the bug in Incomplete state). Because I have been testing 3 different MAAS servers and haven't seen the issue. But now that I look closer at the code, I think this is because there is a separate shim for reporting mDNS entries depending on whether or not the observer is running via the rack or the region. The rack-based observer seems to have the bug, while the region observer is bug-free.

Ultimately, I think this bug would most often be seen in environments with standalone rack controllers.

Looks like this got introduced in MAAS 2.2 due to the getClientNow() cleanup. Previous to that change, the code simply did:

        client = self.clientService.getClient()
        return client(
            ReportMDNSEntries, system_id=client.localIdent, mdns=mdns)