There were 2 reasons for the failing java precommit; there was a consistent compile break that will be fixed by https://github.com/apache/beam/pull/6018 and there was a change, presumably due to the jenkins change, in the number of workers computed based on the number of virtual processors detected which caused the -Xmx gradle jvm flag to decrease from 3g to 1g.
After adjusting maxWorkers (and the memory settings), the precommit passed for pull/6018 at https://builds.apache.org/job/beam_PreCommit_Java_Commit/517/ Setting reasonable values for maxWorkers and Xms/Xmx is being tested in https://github.com/apache/beam/pull/6046 On Tue, Jul 24, 2018 at 9:21 AM Alan Myrvold <amyrv...@google.com> wrote: > I can take a look at it today > > On Mon, Jul 23, 2018 at 10:32 PM Rui Wang <ruw...@google.com> wrote: > >> It is at least frequent (and event consistent) based on the list of >> failed PR which run java checks. >> >> >> Also, this issue is being tracked by >> https://issues.apache.org/jira/browse/BEAM-4847. >> >> >> -Rui >> >> On Mon, Jul 23, 2018 at 9:55 PM Jean-Baptiste Onofré <j...@nanthrax.net> >> wrote: >> >>> Hi Rui >>> >>> I will. I guess it randomly happens, right ? >>> >>> Regards >>> JB >>> Le 24 juil. 2018, à 02:37, Rui Wang <ruw...@google.com> a écrit: >>>> >>>> Hi community, >>>> >>>> Seems like both Java precommit and postcommit tests are failing due to >>>> GC issues. I created a JIRA under test-failures component ( >>>> https://issues.apache.org/jira/browse/BEAM-4848). Could someone who is >>>> familiar with Jenkins take a look at this? >>>> >>>> >>>> Thanks, >>>> Rui >>>> >>>