[
https://issues.apache.org/jira/browse/HDFS-15227?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17060328#comment-17060328
]
Ayush Saxena commented on HDFS-15227:
-------------------------------------
I have put a patch that can (by means of breakpoints) can reproduce the issue.
Put breakpoint at :
{code:java}
//display under construction block info.
if (!blocks.isLastBlockComplete() && lastBlock != null) {
ExtendedBlock block = lastBlock.getBlock(); // First Breakpoint...
{code}
Second here :
{code:java}
@Override
public void run() {
DataNodeTestUtils.resumeIBR(cluster.getDataNodes().get(0)); //Second
here...
{code}
Once the first breakpoint is hit(in the NamenodeFSCK), before progressing
ahead, make the second breakpoint progress to shoot the IBR's.
Then the first breakpoint thread.. This shall throw NPE.
> FSCK -upgradedomains is failing for upgradedomains when more than 2 million
> blocks present in hdfs and write in progress of some blocks
> ---------------------------------------------------------------------------------------------------------------------------------------
>
> Key: HDFS-15227
> URL: https://issues.apache.org/jira/browse/HDFS-15227
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: krishna reddy
> Assignee: Ayush Saxena
> Priority: Major
> Attachments: TestToRepro.patch
>
>
> FSCK -upgradedomains is failing for upgradedomains when more than 2 million
> blocks present in hdfs and write in progress of some blocks
> "hdfs fsck / -files -blocks -upgradedomains"
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
