[
https://issues.apache.org/jira/browse/HDFS-17456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17841864#comment-17841864
]
ASF GitHub Bot commented on HDFS-17456:
---------------------------------------
fuchaohong commented on PR #6713:
URL: https://github.com/apache/hadoop/pull/6713#issuecomment-2081977865
> @fuchaohong Hi, IIUC the `dfsused` will be refreshed periodically. Do you
meet any issues about this metric? Thanks.
@Hexiaoqiao Indeed, the dfsused metric is corrected periodically, but the
cycle is very long. Currently, when there are too many append operations,
dfsused may expand abnormally, not reflecting the actual usage accurately.
Thanks.
> Fix the dfsused statistics of datanode are incorrect when appending a file.
> ---------------------------------------------------------------------------
>
> Key: HDFS-17456
> URL: https://issues.apache.org/jira/browse/HDFS-17456
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs
> Affects Versions: 3.3.3
> Reporter: fuchaohong
> Priority: Major
> Labels: pull-request-available
>
> In our production env, the namenode page showed that the datanode space had
> been used up, but the actual datanode machine still had a lot of free space.
> After troubleshooting, the dfsused statistics of datanode are incorrect when
> appending a file. The following is the dfsused after each append of 100.
> |*Error*|*Expect*|
> |0|0|
> |100|100|
> |300|200|
> |600|300|
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
