hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stefanlee (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-7695) when active RM transit to standby , this RM will new another FairSchedulerUpdate Thread
Date Wed, 03 Jan 2018 08:23:02 GMT

    [ https://issues.apache.org/jira/browse/YARN-7695?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16309278#comment-16309278
] 

stefanlee commented on YARN-7695:
---------------------------------

there is another problem in this scenario, when i turn on *ContinuousScheduling* , and submit
a lot of applications, then my cluster have no available resource,  active RM1's log print

{code:java}
2018-01-03 16:05:49,860 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.AppSchedulable:
Making reservation: node=datanode2 app_id=application_1514952157240_0019
2018-01-03 16:05:49,860 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerApplicationAttempt:
Application application_1514952157240_0019 reserved container container_1514952157240_0019_02_000003
on node host: datanode2:37528 #containers=2 available=<memory:2048, vCores:1> used=<memory:4096,
vCores:2>, currently has 1 at priority 10; currentReservation 3072
2018-01-03 16:05:49,860 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FSSchedulerNode:
Updated reserved container container_1514952157240_0019_02_000003 on node host: datanode2:37528
#containers=2 available=<memory:2048, vCores:1> used=<memory:4096, vCores:2> for
application org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FSSchedulerApp@a9790a8
2018-01-03 16:05:49,868 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler:
Trying to fulfill reservation for application appattempt_1514952157240_0019_000002 on node:
host: datanode2:37528 #containers=2 available=<memory:2048, vCores:1> used=<memory:4096,
vCores:2>
2018-01-03 16:05:49,868 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.AppSchedulable:
Making reservation: node=datanode2 app_id=application_1514952157240_0019
2018-01-03 16:05:49,868 INFO org.apache.hadoop.yarn.server.resourcemanager.scheduler.SchedulerApplicationAttempt:
Application application_1514952157240_0019 reserved container container_1514952157240_0019_02_000003
on node host: datanode2:37528 #containers=2 available=<memory:2048, vCores:1> used=<memory:4096,
vCores:2>, currently has 1 at priority 10; currentReservation 3072
{code}
, then i repeat 3 step in description, active RM1 transit to standby, RM2 runs normaly, but
standby RM1's log still print above info, it seems *ContinuousScheduling*  runs in dead loop.
so  i think it is best to stop these thread when RM1 transit to standby in this  scenario.

> when active RM transit to standby , this RM will new another FairSchedulerUpdate Thread
> ---------------------------------------------------------------------------------------
>
>                 Key: YARN-7695
>                 URL: https://issues.apache.org/jira/browse/YARN-7695
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: fairscheduler, resourcemanager
>    Affects Versions: 2.4.0
>            Reporter: stefanlee
>
> 1. i test haoop-2.4.0 in my cluster.
> 2. RM1 is active and  RM2 is standby
> 3. i delete /yarn-leader-election/Yarn/ActiveStandbyElectorLock from ZK
> 4. RM1 then transit from active to standby success.
> 5. at last ,i print RM1 jstack info and found two "AllocationFileReloader" and two "FairSchedulerUpdateThread"
in RM1.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org


Mime
View raw message