[
https://issues.apache.org/jira/browse/HDFS-8439?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14555505#comment-14555505
]
Liu Shaohui commented on HDFS-8439:
-----------------------------------
[~stack]
{quote}
sometimes you log at INFO level and other times WARN. Is there a reason for
that? If a mistake
{quote}
Sorry, it's my mistake. Fix it in v3 patch.
We have pushed this to production and it works as expected.
> Adding more slow action log in critical read path
> -------------------------------------------------
>
> Key: HDFS-8439
> URL: https://issues.apache.org/jira/browse/HDFS-8439
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Liu Shaohui
> Assignee: Liu Shaohui
> Priority: Minor
> Attachments: HDFS-8439.001.patch, HDFS-8439.002.patch,
> HDFS-8439.003.patch
>
>
> To dig a HBase read spike issue, we'd better to add more slow pread/seek log
> in read flow to get the abnormal datanodes.
> Patch will be uploaded soon.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
