[
https://issues.apache.org/jira/browse/SLIDER-943?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14933089#comment-14933089
]
Steve Loughran commented on SLIDER-943:
---------------------------------------
The check that is failing is a scan through all requests at that priority
group, which is detecting that a request has been issued with the relax flag
set into a group which has an outstanding relax=false.
This didn't surface in the mock tests; they may lack a validation phase which
ought to be pulled in so that we can replicate this in a unit test, then fix.
Otherwise this can only be replicated by trying to deploy onto a cluster
without enough capacity then waiting for things to time out.
> Container Escalation failing
> ----------------------------
>
> Key: SLIDER-943
> URL: https://issues.apache.org/jira/browse/SLIDER-943
> Project: Slider
> Issue Type: Bug
> Components: appmaster
> Affects Versions: Slider 0.80
> Environment: real YARN Cluster
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Priority: Critical
>
> {Code}
> 2015-09-26 18:25:05,533 [AmExecutor-006] ERROR actions.QueueExecutor -
> Exception processing
> org.apache.slider.server.appmaster.actions.RenewingAction@475c6ebf
> name='renewing EscalateOutstandingRequests', delay=0, attrs=0,
> sequenceNumber=5}:
> org.apache.hadoop.yarn.client.api.InvalidContainerRequestException: Cannot
> submit a ContainerRequest asking for location * with locality relaxation
> true when it has already been requested with locality relaxation false^M
> org.apache.hadoop.yarn.client.api.InvalidContainerRequestException: Cannot
> submit a ContainerRequest asking for location * with locality relaxation true
> when it has already been requested with locality relaxation false^M
> at
> org.apache.hadoop.yarn.client.api.impl.AMRMClientImpl.checkLocalityRelaxationConflict(AMRMClientImpl.java:582)^M
> at
> org.apache.hadoop.yarn.client.api.impl.AMRMClientImpl.addContainerRequest(AMRMClientImpl.java:415)^M
> at
> org.apache.hadoop.yarn.client.api.async.impl.AMRMClientAsyncImpl.addContainerRequest(AMRMClientAsyncImpl.java:166)^M
> at
> org.apache.slider.server.appmaster.operations.AsyncRMOperationHandler.addContainerRequest(AsyncRMOperationHandler.java:106)^M
> at
> org.apache.slider.server.appmaster.operations.ContainerRequestOperation.execute(ContainerRequestOperation.java:38)^M
> at
> org.apache.slider.server.appmaster.operations.RMOperationHandler.execute(RMOperationHandler.java:28)^M
> at
> org.apache.slider.server.appmaster.SliderAppMaster.execute(SliderAppMaster.java:1889)^M
> at
> org.apache.slider.server.appmaster.SliderAppMaster.escalateOutstandingRequests(SliderAppMaster.java:1824)^M
> at
> org.apache.slider.server.appmaster.actions.EscalateOutstandingRequests.execute(EscalateOutstandingRequests.java:43)^M
> at
> org.apache.slider.server.appmaster.actions.RenewingAction.execute(RenewingAction.java:88)^M
> at
> org.apache.slider.server.appmaster.actions.QueueExecutor.run(QueueExecutor.java:73)^M
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)^M
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)^M
> at java.lang.Thread.run(Thread.java:745)^M
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
