Comment 11 for bug 1834190

Revision history for this message
Ryan Harper (raharper) wrote : Re: Modules steps not taken into account since 19.1

I cannot recreate the failure with your config drive and the current disco or cosmic image, however it does allow you to login before cloud-init is complete. You can watch/wait for cloud-init to complete (cloud-init status --long --wait). And systemctl list-jobs (will show you any pending units that are not yet complete. To see the extra time it takes look at this:

% systemd-analyze critical-chain cloud-final.service
cloud-final.service +481ms
└─cloud-config.service @1min 10.653s +5.742s
  └─snapd.seeded.service @23.132s +47.511s
    └─snapd.service @54.339s +3.108s
      └─snap-lxd-10972.mount @58.097s +28ms
        └─dev-loop1.device @58.116s +47ms

47 seconds for the snapd.seeded.service to install the two snaps. And the cloud-final.service will wait until the snapd.seeded.sevice is complete as cloud-init defers package and snap installs until after snapd is up.

On your failing instance log, I see a snapd message about error installing the core snap. Could you run:

snap list
snap changes

And for any errors in the snap changes output, please run:

snap tasks <ID> to get an error message.