[
https://issues.apache.org/jira/browse/HBASE-20036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16376468#comment-16376468
]
ramkrishna.s.vasudevan commented on HBASE-20036:
------------------------------------------------
The reason for the failure was that the 6th block may take some time by the
time the assertion happened. So now ensured that we wait for the sixth block
and then do the assert. Hopefully this test is not flaky anymore.
> TestAvoidCellReferencesIntoShippedBlocks timed out
> --------------------------------------------------
>
> Key: HBASE-20036
> URL: https://issues.apache.org/jira/browse/HBASE-20036
> Project: HBase
> Issue Type: Sub-task
> Reporter: stack
> Priority: Major
> Attachments: HBASE-20036.patch
>
>
> Looks like it is stuck can't flush; bad math?
> See the dashboard where it hung here:
> https://builds.apache.org/job/HBASE-Flaky-Tests-branch2.0/2428/artifact/hbase-server/target/surefire-reports/org.apache.hadoop.hbase.client.TestAvoidCellReferencesIntoShippedBlocks-output.txt
> ... the provocation could be this?
> 2018-02-21 04:18:44,973 DEBUG [Thread-178] bucket.BucketCache(629): This
> block eefaa9a7b10e437b9fc2b55a67d63191_4356 is still referred by 1 readers.
> Can not be freed now. Hence will mark this for evicting at a later point
> Exception in thread "Thread-178" java.lang.AssertionError: old blocks should
> still be found expected:<6> but was:<5>
> at org.junit.Assert.fail(Assert.java:88)
> at org.junit.Assert.failNotEquals(Assert.java:834)
> at org.junit.Assert.assertEquals(Assert.java:645)
> ... Then we get stuck doing this:
> 2018-02-21 04:23:34,661 DEBUG [master/asf903:0.Chore.1] master.HMaster(1524):
> Skipping normalization for table: testHBase16372InCompactionWritePath, as
> it's either system table or doesn't have auto normalization turned on
> 2018-02-21 04:23:35,695 INFO [regionserver/asf903:0.Chore.1]
> regionserver.HRegionServer$PeriodicMemStoreFlusher(1752):
> MemstoreFlusherChore requesting flush of hbase:meta,,1.1588230740 because
> info has an old edit so flush to free WALs after random delay 286820ms
> 2018-02-21 04:23:36,009 DEBUG [ReadOnlyZKClient-localhost:61855@0x227ea3ed]
> zookeeper.ReadOnlyZKClient(316): 0x227ea3ed to localhost:61855 inactive for
> 60000ms; closing (Will reconnect when new requests)
> It also failed a recent nightly for same reason:
> https://builds.apache.org/job/HBase%20Nightly/job/branch-2/355/
> Any chance you'd take a look [~ram_krish]? You best at this stuff?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
