OpenStack Autopilot Deployment Error – Fix ‘Failed Status’ on KVM

cloudlandscapeopenstackopenstack-autopilot

after to run the install of openstack-autopilot,
some vnode went in "failed" status some others are in deployed and on Landscape I saw the progress of the deploy for the rest

one question is: "it's normal that?"
second question: "is there a way to re-deploy the vnodes in "failed" status?"

enter image description here

at the end after some minutes I've received that

enter image description here

Best Answer

Just some clarification here

  1. Canonical doesn't recommend or support installing onto KVM nodes with the Landscape Autopilot. If you want to do that, please use the single machine option of openstack-install.
  2. To help you further, more data is required about what failed here. I'd recommend filing a bug with the link that appears after it's done with the 'collecting logs' in your second screenshot. Through that the developers can see what failed a bit more easily.
  3. There is no way to restart just a few failed nodes. It's all or nothing.
  4. It's not normal or expected to hit errors, but it certainly happens. It's a very complex system and many things can go wrong during the deployment, but the goal is for it to work every time or explain why it failed.

If you want to dig in more yourself, get access to the openstack environment, and check the Juju unit logs that failed. The logs should have more data about what failed rather than just 'config_changed hook failed'.

juju ssh UNIT-NAME
cd /var/log/juju
ls unit-*

See my other answer for more details about getting access to the OpenStack Juju environment.

Related Question