[
https://issues.apache.org/jira/browse/YARN-4308?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sunil G updated YARN-4308:
--------------------------
Attachment: 0008-YARN-4308.patch
Thanks [~templedf] for pointing out java doc issue.
Handled all comments and I have some points to add to the sleep issue which you
mentioned.
{{MonitoringThread}} iterates through the process tree of all running
containers and sets its utilization. It also checks for memory over utilization
and kills such containers. However in our current scenario, we are verifying
only CPU values. I also wanted to avoid these sleeps in first place and wanted
to verify based on some events or processed values. However we have only
containerResourceUtilization value to look for as a change and a default value
of {{ResourceUtilization.newInstance(0, 0, 0.0f)}} is set already. So if CPU
readings are coming as 0, this will be still 0. Hence I can do this check only
for the test case which I added as CPU value of 50 was returned by
{{MockCPUResourceCalculatorProcessTree}}.
We can see whether it can be generalized for similar cases in future.
> ContainersAggregated CPU resource utilization reports negative usage in first
> few heartbeats
> --------------------------------------------------------------------------------------------
>
> Key: YARN-4308
> URL: https://issues.apache.org/jira/browse/YARN-4308
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 2.7.1
> Reporter: Sunil G
> Assignee: Sunil G
> Attachments: 0001-YARN-4308.patch, 0002-YARN-4308.patch,
> 0003-YARN-4308.patch, 0004-YARN-4308.patch, 0005-YARN-4308.patch,
> 0006-YARN-4308.patch, 0007-YARN-4308.patch, 0008-YARN-4308.patch
>
>
> NodeManager reports ContainerAggregated CPU resource utilization as -ve value
> in first few heartbeats cycles. I added a new debug print and received below
> values from heartbeats.
> {noformat}
> INFO
> org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl:
> ContainersResource Utilization : CpuTrackerUsagePercent : -1.0
> INFO
> org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl:ContainersResource
> Utilization : CpuTrackerUsagePercent : 198.94598
> {noformat}
> Its better we send 0 as CPU usage rather than sending a negative values in
> heartbeats eventhough its happening in only first few heartbeats.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
