[openstack-dev] [qa] Plan for failing "successful" tempest jobs when new ERRORs appear in logs
mtreinish at kortar.org
Mon Nov 25 19:55:30 UTC 2013
On Mon, Nov 25, 2013 at 11:06:36AM -0800, Joe Gordon wrote:
> On Mon, Nov 18, 2013 at 2:58 PM, David Kranz <dkranz at redhat.com> wrote:
> > So we are close to being able to start doing this. The current whitelist
> > is here https://github.com/openstack/tempest/blob/master/etc/
> > whitelist.yaml. I have a "find-errors" script that watches for successful
> > builds and pulls out the non-whitelisted errors. For the past few weeks I
> > have been doing the following:
> > 1. Run find-errors
> > 2. File bugs on any new errors
> > 3. Add to whitelist
> > 4. Repeat
> > There are still some very flaky cases. I will do one more iteration of
> > this. Right now this script https://github.com/openstack/
> > tempest/blob/master/tools/check_logs.py dumps non-whitelisted errors to
> > the console log but
> > always returns success. The question now is how long should all jobs run
> > with no new errors showing, before changing check_logs.py to fail if there
> > are any new errors?
> The sooner the better.
I would just turn it on today. This is the week to do it because of the holiday.
Based on my experience with flipping the switch for parallel the only way to iron
out all of the kinks is to make it gating so people will notice when something
fails. There will be some pain at first but the end result makes it worth it.
More information about the OpenStack-dev