[
https://issues.apache.org/jira/browse/SOLR-12040?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16795952#comment-16795952
]
Kevin Risden edited comment on SOLR-12040 at 3/19/19 10:27 AM:
---------------------------------------------------------------
While I was looking at SOLR-13330 (and related HDFS tests), I think I saw this
fail due to a race condition shutting down ZK client. I'll have to check again
but the gist was:
* connection manager shutdown
* shutdown ZK client
* zk client stuck in trying to send / read packet during shutdown
* suite eventually times out from not being able to make any progress.
was (Author: risdenk):
While I was looking at SOLR-13330 (and related HDFS tests), I think I saw this
fail due to a race condition shutting down ZK client. I'll have to check again
but the gist was:
* connection manager shutdown
* shutdown ZK client
* zk client stuck in trying to send / read packet during shutdown
* suite eventually times out from not being able to make any progress.
> HdfsBasicDistributedZkTest & HdfsBasicDistributedZk2 fail on virtually every
> jenkins run
> ----------------------------------------------------------------------------------------
>
> Key: SOLR-12040
> URL: https://issues.apache.org/jira/browse/SOLR-12040
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: Hadoop Integration, hdfs, Tests
> Reporter: Hoss Man
> Assignee: Mark Miller
> Priority: Major
>
> HdfsBasicDistributedZkTest & HdfsBasicDistributedZk2 are thin subclasses of
> BasicDistributedZkTest & BasicDistributedZk2 that just tweak the setup to use
> HDFS, and only run @Nightly.
> These tests are failing virtually every time they are run by jenkins - either
> at a method level, or at a suite level (due to threadleaks, timeouts, etc...)
> yet their non-HDFS superclasss virtually never fail.
> Per the jenkins failure rates reports i've setup, here's the failure rates of
> all tests matching "BasicDistributed" for the past 7days (note that the
> non-HDFS tests aren't even listed, because they haven't failed at all even
> though they are non-nightly and have cumulatively run ~750 times in the past
> 7 days)
> http://fucit.org/solr-jenkins-reports/failure-report.html
> {noformat}
> "Suite?","Class","Method","Rate","Runs","Fails"
> "true","org.apache.solr.cloud.hdfs.HdfsBasicDistributedZk2Test","","53.3333333333333","15","8"
> "false","org.apache.solr.cloud.hdfs.HdfsBasicDistributedZk2Test","test","18.75","16","3"
> "true","org.apache.solr.cloud.hdfs.HdfsBasicDistributedZkTest","","46.1538461538462","13","6"
> "false","org.apache.solr.cloud.hdfs.HdfsBasicDistributedZkTest","test","7.69230769230769","13","1"
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
