[
https://issues.apache.org/jira/browse/YARN-2641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14168564#comment-14168564
]
zhihai xu commented on YARN-2641:
---------------------------------
[~kasha] thanks to review the patch.
Just synchronized hostsReader in NodeListManager#isValidNode is not enough.
There is a race condition between ResourceTrackerService#registerNodeManager
and NodeListManager#refreshNodes:
If NodeListManager#refreshNodes happens right after NodeListManager#isValidNode
and before create a new RMNode in ResourceTrackerService#registerNodeManager.
The node to be decommissioned will be added after NodeListManager#refreshNodes.
And it will never be decommissioned until next time
NodeListManager#refreshNodes is called.
By synchronizing isValidNode and create-a-new-RMNode on hostsReader in
ResourceTrackerService#registerNodeManager,
we can make sure the NodeListManager#refreshNodes is called either before
isValidNode or after create-a-new-RMNode.
If it is called before check the node, isValidNode will return false which will
shutdown the node.
If it is called after RMNode is created, the new created RMNode will be
decommissioned by NodeListManager#refreshNodes immediately.
> improve node decommission latency in RM.
> ----------------------------------------
>
> Key: YARN-2641
> URL: https://issues.apache.org/jira/browse/YARN-2641
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: resourcemanager
> Affects Versions: 2.5.0
> Reporter: zhihai xu
> Assignee: zhihai xu
> Attachments: YARN-2641.000.patch, YARN-2641.001.patch
>
>
> improve node decommission latency in RM.
> Currently the node decommission only happened after RM received nodeHeartbeat
> from the Node Manager. The node heartbeat interval is configurable. The
> default value is 1 second.
> It will be better to do the decommission during RM Refresh(NodesListManager)
> instead of nodeHeartbeat(ResourceTrackerService).
> This will be a much more serious issue:
> After RM is refreshed (refreshNodes), If the NM to be decommissioned is
> killed before NM sent heartbeat to RM. The RMNode will never be
> decommissioned in RM. The RMNode will only expire in RM after
> "yarn.nm.liveness-monitor.expiry-interval-ms"(default value 10 minutes) time.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
