Comment 2 for bug 2065734

Revision history for this message
Ethan Myers (ethanmyers) wrote :

Hi Thorsten,

In my case, the networking config didn't match what the switch config was expecting, so the machines never had a valid LAN connection, preventing them from reaching the MAAS server. To reproduce, I expect you could configure the deployed machine with a bad gateway, bad bond, etc to keep the deployed machine from contacting MAAS.

In my case, I was able to successfully commission a node. I tried to deploy that node with a bad networking config, moved it into rescue mode, then was unable to get it to leave rescue mode.

What I want is a timeout on the MAAS side to pickup these kinds of failures. MAAS can't contact the deployed node, so it endlessly tries to leave rescue mode. In my opinion, after 1800 seconds (or whatever timeout you think it appropriate), MAAS gives up trying to leave rescue mode and moves the machine into a FAILED_RELEASING state, or similar.

1. https://discourse-maas-io-uploads.s3.us-east-1.amazonaws.com/original/2X/b/bd9e5e225ffee4b2e88104e5bbd363dd2ef61a88.jpeg