[
https://issues.apache.org/jira/browse/HDDS-401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16622727#comment-16622727
]
Ajay Kumar commented on HDDS-401:
---------------------------------
[~GeLiXin] thanks for the patch. Overall patch looks good. Few comments and
questions:
TestDeadNodeHandler
* testStatisticsUpdate: Shall we emit the actual {{SCMEvents.DEAD_NODE}} event
for datanode1 (L181). This will require EventQueue field in test class and
registration of deadNodeHandler as handler for {{SCMEvents.DEAD_NODE}} event
in setup function. {code} .addHandler(SCMEvents.DEAD_NODE,
deadNodeHandler);{code}
SCMNodeManager#processDeadNode
Not sure what is right answer to this but wanted to raise this as a question.
With current approach if somebody tries to get NodeStat for a dead node, they
will get a NodeNotFoundException. Which may imply node doesn't exist in
cluster. A dead datanode is not decommissioned or removed from cluster. So that
assumption is not correct. Wondering if setting stats for dead node to 0 is
better than removing its entry all together.
> Update storage statistics on dead node
> ---------------------------------------
>
> Key: HDDS-401
> URL: https://issues.apache.org/jira/browse/HDDS-401
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Reporter: Hanisha Koneru
> Assignee: LiXin Ge
> Priority: Major
> Attachments: HDDS-401.000.patch, HDDS-401.001.patch,
> HDDS-401.002.patch, HDDS-401.003.patch
>
>
> This is a follow-up Jira for HDDS-369.
> As per [~ajayydv]'s
> [comment|https://issues.apache.org/jira/browse/HDDS-369?focusedCommentId=16594120&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16594120],
> on detecting a dead node in the cluster, we should update the storage stats
> such as usage, space left.
--
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
