> On Sep 29, 2020, at 3:56 AM, Akira Ajisaka <aajis...@apache.org> wrote:
>
> I'm testing YETUS-994 with a token with repo:status write access,
> however, Yetus couldn't update the commit status as expected.
> https://github.com/apache/hadoop/pull/2348
Try a patch with failures. A status is only put if there are warnings or
failures. If the patch is good, then there is an assumption that the CI system
will also be putting up a status (which, in this case, it did). We can always
change the code to always put a good status, but keep in mind that there are
limits as to how many statuses (statusii?) one is allowed to have...
>
>> Also discovered this morning that Jenkins doesn't always set the GIT_COMMIT
>> env variable which will mess things up too.
>
> Is this the root cause of the failure?
Nope. Doesn't look like there was a failure for it to report. There is
actually a message if the SHA is missing. YETUS-1006 now has a PR which should
fix this problem btw.
>
>> Maybe test-patch should complain if it has a token that lacks this
>> permission?
>
> AFAIK, GitHub returns 404 if the token lacks the permission, so
> test-patch cannot complain about the permission of the token.
You can also get a 404 for other reasons though. :(
