[
https://issues.apache.org/jira/browse/HDFS-14400?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16807066#comment-16807066
]
Karthik Palanisamy commented on HDFS-14400:
-------------------------------------------
[~goiri] Yes, ExpiredHeartbeats is just a counter for dead datanode.
ExpiredHeartbeats is not used by any other functions, it only exposed for
metrics.
{quote}The problem would be if we double counted.
{quote}
It will not be double counted. Assume, One datanode is marked as dead. We
incremented expiredHeartbeat counter to 1, When same datanode node is alive
again after sometime, the counter is not decremented to 0.
The fix is to handle this scenario.
> Namenode ExpiredHeartbeats metric
> ---------------------------------
>
> Key: HDFS-14400
> URL: https://issues.apache.org/jira/browse/HDFS-14400
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs
> Affects Versions: 3.1.2
> Reporter: Karthik Palanisamy
> Assignee: Karthik Palanisamy
> Priority: Minor
> Attachments: HDFS-14400-001.patch, HDFS-14400-002.patch
>
>
> Noticed incorrect value in ExpiredHeartbeats metrics under namenode JMX.
> We will increment ExpiredHeartbeats count when Datanode is dead but somehow
> we missed to decrement when datanode is alive back.
> {code}
> { "name" : "Hadoop:service=NameNode,name=FSNamesystem", "modelerType" :
> "FSNamesystem", "tag.Context" : "dfs", "tag.TotalSyncTimes" : "7 ",
> "tag.HAState" : "active", ... "ExpiredHeartbeats" : 2, ... }
> {code}
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
