On Wed, 15 Sept 2021 at 13:16, Daniel P. Berrangé <berra...@redhat.com> wrote: > > On Wed, Sep 15, 2021 at 01:12:23PM +0100, Peter Maydell wrote: > > On Wed, 15 Sept 2021 at 09:29, Daniel P. Berrangé <berra...@redhat.com> > > wrote: > > > Urgh, my bad, I completely forget this behaviour when reviewing. > > > When we only have > > > > > > when: manual > > > > > > then the job has to be manually started, and it still contributes > > > to pipeline status, so it /must/ triggered manually. > > > > > > If we want it to be manually started and not contribute to the > > > pipeline status we need: > > > > > > rules: > > > ... > > > when: manual > > > allow_failure: true > > > > So there's no way to say "if it is triggered, then it must > > not fail, but if it is not triggered, that's OK" ? > > Not that I've found. > > > I guess it's not a big deal either way though. > > So the fix is to add back the allow_failure tag to those jobs > > which are manual. I'll send a patch... > > Note "allow_failure" is allowed both at the top level of the job > and inside the "rules:". I find it clearer if we put it against > the "rules:" section as shown above.
OK, I guess. That does mean that we end up with two allow_failure lines for each of these jobs, though, beacuse they have two separate if clauses in their rules. -- PMM
