[
https://issues.apache.org/jira/browse/HDFS-11671?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andrew Wang updated HDFS-11671:
-------------------------------
Attachment: HDFS-11671.001.patch
This was a fun one to trace down. I had a pretty reliable repro on our internal
jenkins boxes hosted on EC2. Turns out, the monotime values were very small,
less than the heartbeat expiry, meaning that setting the DN's last seen
monotime to 0 was insufficient to mark it as dead.
I also added some additional guard rails to the test, since I noticed that
sometimes one of the DataStreamers fails to close, and we need to trigger block
replication calculation before checking the queues the first time.
> TestReconstructStripedBlocks#test2RecoveryTasksForSameBlockGroup fails
> ----------------------------------------------------------------------
>
> Key: HDFS-11671
> URL: https://issues.apache.org/jira/browse/HDFS-11671
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: erasure-coding, test
> Affects Versions: 3.0.0-alpha2
> Reporter: Andrew Wang
> Attachments: HDFS-11671.001.patch
>
>
> This failed on a unit test run with 3.0.0-alpha2.
> {noformat}
> java.lang.AssertionError: expected:<1> but was:<0>
> at org.junit.Assert.fail(Assert.java:88)
> at org.junit.Assert.failNotEquals(Assert.java:743)
> at org.junit.Assert.assertEquals(Assert.java:118)
> at org.junit.Assert.assertEquals(Assert.java:555)
> at org.junit.Assert.assertEquals(Assert.java:542)
> at
> org.apache.hadoop.hdfs.server.namenode.TestReconstructStripedBlocks.test2RecoveryTasksForSameBlockGroup(TestReconstructStripedBlocks.java:223)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
