[
https://issues.apache.org/jira/browse/HDFS-10730?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yiqun Lin updated HDFS-10730:
-----------------------------
Description:
In HDFS-10723, [~kihwal] suggested that
{quote}
it is not a good idea to hard-core or reuse the same port number in unit tests.
Because the jenkins slave can run multiple jobs at the same time.
{quote}
Then I collected some tests which failed by this reason in recent jenkin
buildings.
Finally I found these two failed test
{{TestFileChecksum.testStripedFileChecksumWithMissedDataBlocks1}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16301/testReport/)
and
{{TestDecommissionWithStriped.testDecommissionWithURBlockForSameBlockGroup}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16257/testReport/).
We can make a change to update the param value for {{keepPort}} from
{code}
cluster.restartDataNode(dnp, true);
{code}
to
{code}
cluster.restartDataNode(dnp, false);
{code}
was:
In HDFS-10723, [~kihwal] suggested that
{quote}
it is not a good idea to hard-core or reuse the same port number in unit tests.
Because the jenkins slave can run multiple jobs at the same time.
{quote}
So I collected some tests which failed by this reason in recent jenkin
buildings.
Finally I found these two failed test
{{TestFileChecksum.testStripedFileChecksumWithMissedDataBlocks1}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16301/testReport/)
and
{{TestDecommissionWithStriped.testDecommissionWithURBlockForSameBlockGroup}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16257/testReport/).
We can make a change to update the param value for {{keepPort}} from
{code}
cluster.restartDataNode(dnp, true);
{code}
to
{code}
cluster.restartDataNode(dnp, false);
{code}
> Fix some failed tests due to BindException
> ------------------------------------------
>
> Key: HDFS-10730
> URL: https://issues.apache.org/jira/browse/HDFS-10730
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Yiqun Lin
> Assignee: Yiqun Lin
> Attachments: HDFS-10730.001.patch
>
>
> In HDFS-10723, [~kihwal] suggested that
> {quote}
> it is not a good idea to hard-core or reuse the same port number in unit
> tests. Because the jenkins slave can run multiple jobs at the same time.
> {quote}
> Then I collected some tests which failed by this reason in recent jenkin
> buildings.
> Finally I found these two failed test
> {{TestFileChecksum.testStripedFileChecksumWithMissedDataBlocks1}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16301/testReport/)
> and
> {{TestDecommissionWithStriped.testDecommissionWithURBlockForSameBlockGroup}}(https://builds.apache.org/job/PreCommit-HDFS-Build/16257/testReport/).
> We can make a change to update the param value for {{keepPort}} from
> {code}
> cluster.restartDataNode(dnp, true);
> {code}
> to
> {code}
> cluster.restartDataNode(dnp, false);
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
