Open Stack

On Thu, Jul 24, 2014 at 04:31:05PM +0000, Jeremy Stanley wrote:
> On 2014-07-18 15:09:34 +0100 (+0100), Daniel P. Berrange wrote:
> [...]
> > If there were multiple failures and only some were identified, it would
> > be reasonable to *not* automatically recheck.
> [...]
> 
> Another major blocker is that we often add signatures for failures
> which occur 100% of the time, and while those tend to get fixed a
> bit faster than 1% failures, automatic rechecks would mean that for
> some period while we're investigating the gate would just be
> spinning over and over running jobs which had no chance of passing.
> 
> I suppose it could be argued that elastic-recheck needs a
> categorization mechanism so that it also won't recommend rechecking
> for those sorts of scenarios (all discussion of automated rechecks
> aside).

Yep, if there's a bug which is known to hit 90%+ of the time due
to some known problem, launchpad could be tagged with "NoRecheck"
and e-r taught to avoid re-queuing such failures.

Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|