IMHO, most of the problems with ASF Jenkins come down to not having
anyone who actually manages it.
After looking over the list of jobs that infra shared, there’s a ton of
cleanup that really needs to happen. The number of obviously abandoned jobs
but still configured in Jenkins is remarkably high. (e.g., Whirr has been in
the attic for 3 years now...) Removing those jobs would speed up a lot of
things as well as remove a lot of the resource exhaustion.
I think BigTop got this correct. Many projects might be better off with
private Jenkins instances privately funded and managed, but shared with their
community. When/if that community dies, so do those resources with probably no
real impact on the other Apache communities. Plus it avoids the obvious
problems that the shared ASF resources clearly have.
> On Apr 25, 2018, at 4:51 PM, Sean Busbey <[email protected]> wrote:
>
> looks like another build earlier that day on the same host hung for 12 hours:
>
> https://builds.apache.org/job/PreCommit-Admin/13059/
>
> Both jobs look to be failing during the jenkins git plugin due to
> resource exhaustion. IIRC, that plugin is before the timeout starts
> counting.
>
> I suspect if we switched this build over to either manually doing the
> git actions in our job, or moved to the jenkins pipeline DSL we could
> have the timeout handle git operations as well. Wouldn't really help
> with having e.g. H32 be bad.
>
> On Tue, Apr 24, 2018 at 11:09 PM, 张铎(Duo Zhang) <[email protected]> wrote:
>> Seems to the wrong mailing-list...
>>
>> 2018-04-25 12:04 GMT+08:00 张铎(Duo Zhang) <[email protected]>:
>>
>>> https://builds.apache.org/job/PreCommit-Admin/13071/
>>>
>>> The machine was broken and we were stuck there for more than 4 hours until
>>> I killed the task manully...
>>>
>>> There is a 5 minutes timeout for the job, it is strange that why it does
>>> not work...
>>>
