> On 18 Aug 2015, at 12:24, Guillermo Polito <[email protected]> wrote:
>
> Some monkey random questions:
>
> - I saw the monkey failing validations because of internal errors like
> timeouts while running the lint rules. That makes me think, when the monkey
> fails a validation because of a monkey failure, shouldn't it be better to
> mark the issue as "Resolved (Fix Review Needed)" again? Like that the monkey
> can re-pass if there was a timeout or whatever.
>
The danger is that we will then have lots of cases of jobs looping… fail,
rerun, fail, rerun….
> Also, if we need to know that the monkey failed we can always add an extra
> tag on it.
>
> Another solution for this is to add the monkey a retry.
>
> - I saw several times that the monkey rechecks already validated issues. Is
> this a kind of race condition? Does anyone know the reason of this? Maybe we
> can add an extra check to avoid this.
>
The idea is that an issue is checked once for every released update. Because
imagine it was green on Friday, we integrated stuff over the weekend that
touched the same code,
now we just integrate the friday change —> build wlll fail due to merge or
tests will fail, huge mess.
Marcus
