The test went fine. The buildbot now always deletes its experiments after it is
done. In case of a regression, it stores the full experiment directory in a
tarball inside the docker volume. The infrastructure repository contains a
script to fetch these tarballs from the volume and unpack them, and the buildbot
will tell you how to do this.
We are currently ignoring some attributes like total_time in the regression
tests because only small instances are tested and their values are below the
reporting threshold. We should change this but it is unrelated to this issue.
I'll add a note to issue871.
Any objections to closing this?
|