[
https://issues.apache.org/jira/browse/HDFS-10341?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Akira AJISAKA updated HDFS-10341:
---------------------------------
Attachment: HDFS-10341.02.patch
Thanks Arpit for the comment. Updated the patch.
I wanted to expose the *total* number of timeouts, so I created additional
AtomicInt in the previous patch. In the 02 patch, the metric shows the
*current* number of timed out pending replication blocks because
{{timedOutItems}} is cleared in {{getTimedOutBlocks()}}.
I suspect that the current number doesn't fit for us because the number is
cleared very frequently (recheck interval is 3 sec by default, which is
probably smaller than the interval of metrics sink). What do you think?
> Add a metric to expose the timeout number of pending replication blocks
> -----------------------------------------------------------------------
>
> Key: HDFS-10341
> URL: https://issues.apache.org/jira/browse/HDFS-10341
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Akira AJISAKA
> Assignee: Akira AJISAKA
> Attachments: HDFS-10341.01.patch, HDFS-10341.02.patch
>
>
> Per HDFS-6682, recording the timeout number of pending replication blocks is
> useful to get the cluster health.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
