[
https://issues.apache.org/jira/browse/HDFS-17129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17783071#comment-17783071
]
ASF GitHub Bot commented on HDFS-17129:
---------------------------------------
yuanboliu commented on PR #6244:
URL: https://github.com/apache/hadoop/pull/6244#issuecomment-1793981740
+1.
We've seen that a block is marked as missing block because of the mis-order
after applying [HDFS-16016](https://issues.apache.org/jira/browse/HDFS-16016).
This bug can be reproduced as below:
1. client write block to dn1 and dn2 as pipeline
2. dn1/dn2 has over 1 million blocks and fbr is reported one by one disk.
3. when fbr is being reported, client finish writing, dn1/dn2 starts to
trigger ibr in different thread.
4. nn receives all the operations and because of the namesystem write lock,
the operations are executed in line as dn1-ibr, dn2-ibr, dn1-fbr-disk,
dn2-fbr-disk. After this, the block is marked as a missing block.
> mis-order of ibr and fbr on datanode
> -------------------------------------
>
> Key: HDFS-17129
> URL: https://issues.apache.org/jira/browse/HDFS-17129
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 3.4.0
> Environment: hdfs3.4.0
> Reporter: liuguanghua
> Assignee: liuguanghua
> Priority: Major
> Labels: pull-request-available
>
> HDFS-16016 , provide new thread to handler IBR. That is a greate improvement.
> But it maybe casue the mis-order of ibr and fbr
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
