[hibernate-dev] Stuck build on CI

Yoann Rodiere yoann at hibernate.org
Mon Jul 24 03:35:21 EDT 2017


> This build on CI seems stuck since yesterday:

Yes, there was a race condition that I fixed before leaving on Friday.
Sorry I didn't realize this build was still running.

> Also I noticed a huge amount of log output: "121,741 KB". Do you really
> need all that output for each build? Esp. given that this job is about
> perf, logging that much seems concerning potentially.

No we don't need that much output, but fortunately we don't usually get
that much output. It's only because JMH interrupted all threads when
benchmarks timed out, which made bulk requests fail, which means the bulk
requests were printed out as part of the stack trace. And since you've got
250 requests per bulk requests, that's a lot.
Not sure it's worth it to tune the test so that we don't get that much
output when failing though, since obviously performance tests shouldn't
fail that hard if I did my work correctly.
I'll add a ticket about adding a special case for logging interrupted
requests, though, because users may very well end up in a similar situation
if they want to stop their application by sending out thread interrupts.


Yoann Rodière
Hibernate NoORM Team
yoann at hibernate.org

On 23 July 2017 at 09:52, Gunnar Morling <gunnar at hibernate.org> wrote:

> Hi Sanne, Yoann,
>
> This build on CI seems stuck since yesterday:
>
>
> http://ci.hibernate.org/job/hibernate-search-performance-elasticsearch/44/
>
> Also I noticed a huge amount of log output: "121,741 KB". Do you really
> need all that output for each build? Esp. given that this job is about
> perf, logging that much seems concerning potentially.
>
> Cheers,
>
> --Gunnar
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>


More information about the hibernate-dev mailing list