Code review comment for ~falcojr/curtin:update-manager-in-docs

> The autoland job just runs a plain 'tox' on the source tree [1], and I
> can reproduce that problem locally on my Groovy system.

Looks like tox.ini needs a tip-flake8 so we can reproduce this elsewhere.

>
> It's annoying when the autolander job runs in this kind of loop. One
> easy way to prevent it would be to have the autolander job to set the
> MP back to "Needs review" on failure. What do you think?

+1 on moving back to needs review.

> It would be nice to have it set the MP back to "Needs review" after e.g. 3
> failures. I was thinking of doing this using the Jenkins "unstable" state, but
> this is not possible as the status of job is not per-MP.
>
> [1] https://github.com/canonical/server-jenkins-jobs/blob/master/curtin/jobs-ci.yaml#L187

So can it be done in our jobs or not?