[
https://issues.apache.org/jira/browse/HDFS-15010?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16983201#comment-16983201
]
Surendra Singh Lilhore commented on HDFS-15010:
-----------------------------------------------
{quote}{{TestFsVolumeList}} looks suspicious.
{quote}
{{addReplicaThreadPool}} is a static instance and if some other test case is
using {{MiniDFSCluster}} then itwill initialize with
{{Runtime.getRuntime().availableProcessors()}}. This will pass in some machine
if available processors in that machine is 5. So it is better to reinitialize
{{addReplicaThreadPool}} before test case execution.
{quote}BTW, I checked the test and is a little long (~5 seconds).
{quote}
Its depend on machine, if you see the latest test report it is taking only 3.8
sec.
[https://builds.apache.org/job/PreCommit-HDFS-Build/28400/testReport/org.apache.hadoop.hdfs.server.datanode.fsdataset.impl/TestFsVolumeList/]
I will attache new patch with the fix.
> BlockPoolSlice#addReplicaThreadPool static pool should be initialized by
> static method
> --------------------------------------------------------------------------------------
>
> Key: HDFS-15010
> URL: https://issues.apache.org/jira/browse/HDFS-15010
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: datanode
> Affects Versions: 3.1.2
> Reporter: Surendra Singh Lilhore
> Assignee: Surendra Singh Lilhore
> Priority: Major
> Attachments: HDFS-15010.001.patch, HDFS-15010.02.patch,
> HDFS-15010.03.patch
>
>
> {{BlockPoolSlice#initializeAddReplicaPool()}} method currently initialize the
> static thread pool instance. But when two {{BPServiceActor}} actor try to
> load block pool parallelly then it may create different instance.
> So {{BlockPoolSlice#initializeAddReplicaPool()}} method should be a static
> method.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
