You'd have to check what actually happened to the forked JVM, Steve. The heartbeat is emitted by the controller JVM; a forked JVM should have terminated long before -- it probably hit a JVM error or something else that prevented normal termination (timeout).
Occasionally we can't even get a stack trace from those zombie JVMs. Dawid On Tue, Feb 2, 2016 at 2:32 PM, Steve Rowe <[email protected]> wrote: > I just killed this job: > <https://builds.apache.org/job/Lucene-Solr-NightlyTests-5.x/1085/> > > Just before I killed it, the job's status was: > > Started 5 days 10 hr ago > Build has been executing for 5 days 10 hr on lucene > > and the last line of the console was: > > [junit4] HEARTBEAT J1 PID(1824@lucene1-us-west): 2016-02-02T13:10:19, stalled > for 460516s at: CollectionsAPIDistributedZkTest.test > > Even though this is a nightly job, I think allowing it to run for more than a > few hours is excessive. > > -- > Steve > www.lucidworks.com > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
