Hang in fetching cloud-init data

[Peter Sabaini]

Symptoms:

Attempting to boot a node into rescue mode, testing or (re)commissioning would fail. Node would PXE-boot until it attempted to get metadata via cloud-init and then would hang, respectively cloud-init on the node would time out after a while.

After restarting maas-rackd and maas-regiond I could commission the node again.

I have unfortunately no way to reproduce this.

Versions:

Ubuntu 18.04.4 LTS
MAAS 2.4.2-7034-g2f5deb8b8-0ubuntu1

[Peter Sabaini (peter-sabaini)]

Fwiw I see bug #1910552 possibly describing something similar

[Peter Sabaini (peter-sabaini)]

Uploaded sosreport here: https://private-fileshare.canonical.com/~sabaini/d135ae4f-c2e0-419f-a9e5-2e5c77d19e98/sosreport-maas-2021-04-21-tjlvcyp.tar.xz (Canonical only, sorry)

[Peter Sabaini (peter-sabaini)]

I'm not sure if this should be marked duplicate.

In my mind there's two issues here:

1. a deployed node needing a functioning maas-regiond to come up

2. maas-regiond failing to respond to cloud-init in the first place