[
https://issues.apache.org/jira/browse/HDFS-11306?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15813865#comment-15813865
]
Yongjun Zhang commented on HDFS-11306:
--------------------------------------
Hi [~jojochuang],
Thanks much for working on this issue!
Some comments of the patch:
# Suggest to print a summary WARN message at the beginning of
{{dumpRemainingEditLogs()}}, stating something like "The edits buffer should
have been flushed but there are still <numTxns> unflushed. Below are the list
of the unflushed transactions:".
# add a finally block and call
{code}
IOUtils.cleanup(LOG, dis);
IOUtils.cleanup(LOG, bis);
{code}
# Can we add couple of more different edits in the test?
Thanks.
> Print remaining edit logs from buffer if edit log can't be rolled.
> ------------------------------------------------------------------
>
> Key: HDFS-11306
> URL: https://issues.apache.org/jira/browse/HDFS-11306
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: ha, namenode
> Reporter: Wei-Chiu Chuang
> Assignee: Wei-Chiu Chuang
> Attachments: HDFS-11306.001.patch
>
>
> In HDFS-10943 [~yzhangal] reported that edit log can not be rolled due to
> unexpected edit logs lingering in the buffer.
> Unable to root cause the bug, I propose that we dump the remaining edit logs
> in the buffer into namenode log, before crashing namenode. Use this new
> capability to find the ops that sneaks into the buffer unexpectedly, and
> hopefully catch the bug.
> This effort is orthogonal, but related to HDFS-11292, which adds additional
> informational logs to help debug this issue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
