You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@beam.apache.org by Brian Hulette <bh...@google.com> on 2019/12/18 21:25:30 UTC
PostCommit_Py_VR_Dataflow timing out
It looks like beam_PostCommit_Py_VR_Dataflow has been timing out at 1h40m
since Dec 4 [1]. Are there any objections to bumping up the timeout to
alleviate this? Or any other thoughts on potential causes and/or solutions?
Brian
[1] https://builds.apache.org/job/beam_PostCommit_Py_VR_Dataflow/
Re: PostCommit_Py_VR_Dataflow timing out
Posted by Udi Meiri <eh...@google.com>.
On Thu, Dec 19, 2019 at 12:06 PM Kenneth Knowles <ke...@apache.org> wrote:
> Is there a default timeout for all individual test methods? Can we make
> longer timeouts opt-in?
>
Long-term: yes, once they're migrated to run via pytest. (
https://pypi.org/project/pytest-timeout/)
>
> On Wed, Dec 18, 2019 at 2:27 PM Brian Hulette <bh...@google.com> wrote:
>
>> Ah thanks for the jira link. There's some critical context there - this
>> seems to be caused by a deadlock, so increasing the timeout won't make more
>> tests finish/pass, it will just consume a jenkins slot for longer.
>>
>> On Wed, Dec 18, 2019 at 1:43 PM Udi Meiri <eh...@google.com> wrote:
>>
>>> Yes, there are objections since this would take up a Jenkins slot for
>>> longer.
>>>
>>> An alternative would be to set timeouts on individual tests.
>>> Debugging options: run the gradle tasks locally, try to pinpoint the
>>> culprit PR
>>>
>>> https://issues.apache.org/jira/browse/BEAM-8877
>>>
>>> On Wed, Dec 18, 2019 at 1:25 PM Brian Hulette <bh...@google.com>
>>> wrote:
>>>
>>>> It looks like beam_PostCommit_Py_VR_Dataflow has been timing out at
>>>> 1h40m since Dec 4 [1]. Are there any objections to bumping up the timeout
>>>> to alleviate this? Or any other thoughts on potential causes and/or
>>>> solutions?
>>>>
>>>> Brian
>>>>
>>>> [1] https://builds.apache.org/job/beam_PostCommit_Py_VR_Dataflow/
>>>>
>>>
Re: PostCommit_Py_VR_Dataflow timing out
Posted by Kenneth Knowles <ke...@apache.org>.
Is there a default timeout for all individual test methods? Can we make
longer timeouts opt-in?
On Wed, Dec 18, 2019 at 2:27 PM Brian Hulette <bh...@google.com> wrote:
> Ah thanks for the jira link. There's some critical context there - this
> seems to be caused by a deadlock, so increasing the timeout won't make more
> tests finish/pass, it will just consume a jenkins slot for longer.
>
> On Wed, Dec 18, 2019 at 1:43 PM Udi Meiri <eh...@google.com> wrote:
>
>> Yes, there are objections since this would take up a Jenkins slot for
>> longer.
>>
>> An alternative would be to set timeouts on individual tests.
>> Debugging options: run the gradle tasks locally, try to pinpoint the
>> culprit PR
>>
>> https://issues.apache.org/jira/browse/BEAM-8877
>>
>> On Wed, Dec 18, 2019 at 1:25 PM Brian Hulette <bh...@google.com>
>> wrote:
>>
>>> It looks like beam_PostCommit_Py_VR_Dataflow has been timing out at
>>> 1h40m since Dec 4 [1]. Are there any objections to bumping up the timeout
>>> to alleviate this? Or any other thoughts on potential causes and/or
>>> solutions?
>>>
>>> Brian
>>>
>>> [1] https://builds.apache.org/job/beam_PostCommit_Py_VR_Dataflow/
>>>
>>
Re: PostCommit_Py_VR_Dataflow timing out
Posted by Brian Hulette <bh...@google.com>.
Ah thanks for the jira link. There's some critical context there - this
seems to be caused by a deadlock, so increasing the timeout won't make more
tests finish/pass, it will just consume a jenkins slot for longer.
On Wed, Dec 18, 2019 at 1:43 PM Udi Meiri <eh...@google.com> wrote:
> Yes, there are objections since this would take up a Jenkins slot for
> longer.
>
> An alternative would be to set timeouts on individual tests.
> Debugging options: run the gradle tasks locally, try to pinpoint the
> culprit PR
>
> https://issues.apache.org/jira/browse/BEAM-8877
>
> On Wed, Dec 18, 2019 at 1:25 PM Brian Hulette <bh...@google.com> wrote:
>
>> It looks like beam_PostCommit_Py_VR_Dataflow has been timing out at 1h40m
>> since Dec 4 [1]. Are there any objections to bumping up the timeout to
>> alleviate this? Or any other thoughts on potential causes and/or solutions?
>>
>> Brian
>>
>> [1] https://builds.apache.org/job/beam_PostCommit_Py_VR_Dataflow/
>>
>
Re: PostCommit_Py_VR_Dataflow timing out
Posted by Udi Meiri <eh...@google.com>.
Yes, there are objections since this would take up a Jenkins slot for
longer.
An alternative would be to set timeouts on individual tests.
Debugging options: run the gradle tasks locally, try to pinpoint the
culprit PR
https://issues.apache.org/jira/browse/BEAM-8877
On Wed, Dec 18, 2019 at 1:25 PM Brian Hulette <bh...@google.com> wrote:
> It looks like beam_PostCommit_Py_VR_Dataflow has been timing out at 1h40m
> since Dec 4 [1]. Are there any objections to bumping up the timeout to
> alleviate this? Or any other thoughts on potential causes and/or solutions?
>
> Brian
>
> [1] https://builds.apache.org/job/beam_PostCommit_Py_VR_Dataflow/
>