[
https://issues.apache.org/jira/browse/HDFS-14669?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16897723#comment-16897723
]
qiang Liu commented on HDFS-14669:
----------------------------------
[~ayushtkn] thanks your review
it make sense to me that we only should make changes this Jira mentioned, so
maybe we should just take the v2 patch [^HDFS-14669-trunk.002.patch], which
only fix the file name issue.
then one thing is left, what to do with the unnecessary recheck logics, I will
list some points below,
# by mention "recheck logics" in this Jira I only mean the recheck logics in
this specific test function, not mean all thease kind of logics introduced by
HDFS-13819 , actually HDFS-13819 solved "java.lang.AssertionError: expected:<2>
but was:<1>" , but it was not caused by this function, because at that time,
this function checks some numbers equal to 0
# left the recheck logics as it is has one little problem and maybe one
advantage
## problem: once the assert failes, produce a missleading exceptin, and more,
swallow the assert exceptin makes event looks into log will got nothing
## advantage: the recheck is not necessary only when serveral condtion match(1
datanode , no lazy persist, only one type of storage, currentlly this test
match), but thease condition may change latter, then recheck may be needed.
while on the other hand, maybe better we add recheck logics when we change the
test minicluster config.
conclusion:
# it's ok we take v2 patch of this Jira
# the recheck logics need some fix, either remove it or move the
"scanner.reconcile();" inter the recheck wait for(more detail see prev comments)
[~elgoiri] [~ayushtkn] any suggestion is welcome
> TestDirectoryScanner#testDirectoryScannerInFederatedCluster fails
> intermittently in trunk
> -----------------------------------------------------------------------------------------
>
> Key: HDFS-14669
> URL: https://issues.apache.org/jira/browse/HDFS-14669
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: datanode
> Affects Versions: 3.2.0
> Environment: env free
> Reporter: qiang Liu
> Assignee: qiang Liu
> Priority: Minor
> Labels: scanner, test
> Attachments: HDFS-14669-trunk-001.patch, HDFS-14669-trunk.002.patch,
> HDFS-14669-trunk.003.patch
>
>
> org.apache.hadoop.hdfs.server.datanode.TestDirectoryScanner#testDirectoryScannerInFederatedCluster
> radomlly Failes because of write files of the same name, meaning intent to
> write 2 files but 2 files are the same name, witch cause a race condition of
> datanode delete block and the scan action count block.
>
> Ref ::
> [https://builds.apache.org/job/hadoop-qbt-trunk-java8-linux-x86/1207/testReport/junit/org.apache.hadoop.hdfs.server.datanode/TestDirectoryScanner/testDirectoryScannerInFederatedCluster/]
>
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
