[
https://issues.apache.org/jira/browse/SLIDER-270?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14086379#comment-14086379
]
Steve Loughran commented on SLIDER-270:
---------------------------------------
Looks like a fifo scheduler issue, or we are sending wrong container requests
in.
Key point: if the containers from the first flex down aren't being noted as
released, we can't find new containers to release the next time round, because
they are still marked as "being released".
...but the root cause is not so clear
> Calling flex (down) the second time does not work
> -------------------------------------------------
>
> Key: SLIDER-270
> URL: https://issues.apache.org/jira/browse/SLIDER-270
> Project: Slider
> Issue Type: Bug
> Components: appmaster
> Affects Versions: Slider 0.50
> Reporter: Sumit Mohanty
> Assignee: Steve Loughran
> Fix For: Slider 0.50
>
>
> From AppMaster log (see below) it looks like that the second command to flex
> from 2 to 1 did not result in container release.
> {noformat}
> 14/08/04 01:55:18 INFO state.AppState: Role MEMCACHED flexed from 3 to 2
> 14/08/04 01:55:18 INFO state.AppState: RoleStatus{name='MEMCACHED', key=1,
> desired=2, actual=3, requested=0, releasing=0, failed=0, started=3,
> startFailed=0, completed=0, failureMessage=''}
> 14/08/04 01:55:18 INFO state.AppState: MEMCACHED: Asking for 1 fewer node(s)
> for a total of 2
> 14/08/04 01:55:19 INFO appmaster.SliderAppMaster: onContainersCompleted([1]
> 14/08/04 01:55:19 INFO appmaster.SliderAppMaster: Container Completion for
> containerID=container_1405048900371_0054_01_000004, state=COMPLETE,
> exitStatus=-100, diagnostics=Container released by application
> 14/08/04 01:55:19 INFO state.AppState: Container was queued for release
> 14/08/04 01:55:19 INFO state.AppState: decrementing role count for role
> MEMCACHED
> 14/08/04 01:55:19 INFO agent.AgentProviderService: Removing container
> specific data for container_1405048900371_0054_01_000004
> 14/08/04 01:55:19 INFO agent.AgentProviderService: publishing
> PublishedConfiguration{description='ComponentInstanceData' entries = 2}
> 14/08/04 01:55:19 INFO state.AppState: RoleStatus{name='MEMCACHED', key=1,
> desired=2, actual=2, requested=0, releasing=0, failed=0, started=3,
> startFailed=0, completed=1, failureMessage=''}
> 14/08/04 01:55:45 INFO state.AppState: Role MEMCACHED flexed from 2 to 1
> 14/08/04 01:55:45 INFO state.AppState: RoleStatus{name='MEMCACHED', key=1,
> desired=1, actual=2, requested=0, releasing=0, failed=0, started=3,
> startFailed=0, completed=1, failureMessage=''}
> 14/08/04 01:55:45 INFO state.AppState: MEMCACHED: Asking for 1 fewer node(s)
> for a total of 1
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
