On Mon, Jul 18, 2016 at 2:09 PM, Andreas Jaeger <aj at suse.com> wrote: > In general, I think coverage jobs should be run as check job so > that you know how the coverage changes. Running them only in the post > job means that in practice nobody sees the output of the job. I look at the post queue coverage output for nova, and have often wished I could see the check queue coverage results. +1 for moving it to the check queue. > Here're the failures (including some that run successful but coverage > > nova: > http://logs.openstack.org/3f/3f700b5a5a498ba08e77378d34f059c3fa6845d8/post/nova-coverage-db/2bdabf8/ I suspect you can cross nova off your list of failures soon. I believe those errors were addressed by the following: https://bugs.launchpad.net/nova/+bug/1603979 Here's a recent passing coverage run for nova (from before those errors were introduced): http://logs.openstack.org/1a/1a074a72845d812666a904b9c17289a43bb32062/post/nova-coverage-db/2fe5506/cover/ Much appreciated, --diana