[
https://issues.apache.org/jira/browse/YARN-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16908908#comment-16908908
]
Bibin A Chundatt commented on YARN-9738:
----------------------------------------
[~sunilg]
Only for the nodes currently its changed to concurrentHashMap will give bucket
level locking.
Also the read lock is removed only for getNodeReport invoked from
ClientRmService,NodeInfo,ApplicationMasterService ,explicitly null check is
handled too.
Patch looks good to me , can we go ahead ??
> Remove lock on ClusterNodeTracker#getNodeReport as it blocks application
> submission
> -----------------------------------------------------------------------------------
>
> Key: YARN-9738
> URL: https://issues.apache.org/jira/browse/YARN-9738
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Bilwa S T
> Assignee: Bilwa S T
> Priority: Major
> Attachments: YARN-9738-001.patch, YARN-9738-002.patch
>
>
> *Env :*
> Server OS :- UBUNTU
> No. of Cluster Node:- 9120 NMs
> Env Mode:- [Secure / Non secure]Secure
> *Preconditions:*
> ~9120 NM's was running
> ~1250 applications was in running state
> 35K applications was in pending state
> *Test Steps:*
> 1. Submit the application from 5 clients, each client 2 threads and total 10
> queues
> 2. Once application submittion increases (for each application of
> distributted shell will call getClusterNodes)
> *ClientRMservice#getClusterNodes tries to get
> ClusterNodeTracker#getNodeReport where map nodes is locked.*
> {quote}
> "IPC Server handler 36 on 45022" #246 daemon prio=5 os_prio=0
> tid=0x00007f75095de000 nid=0x1949c waiting on condition [0x00007f74cff78000]
> java.lang.Thread.State: WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <0x00007f759f6d8858> (a
> java.util.concurrent.locks.ReentrantReadWriteLock$FairSync)
> at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:836)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireShared(AbstractQueuedSynchronizer.java:967)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireShared(AbstractQueuedSynchronizer.java:1283)
> at
> java.util.concurrent.locks.ReentrantReadWriteLock$ReadLock.lock(ReentrantReadWriteLock.java:727)
> at
> org.apache.hadoop.yarn.server.resourcemanager.scheduler.ClusterNodeTracker.getNodeReport(ClusterNodeTracker.java:123)
> at
> org.apache.hadoop.yarn.server.resourcemanager.scheduler.AbstractYarnScheduler.getNodeReport(AbstractYarnScheduler.java:449)
> at
> org.apache.hadoop.yarn.server.resourcemanager.ClientRMService.createNodeReports(ClientRMService.java:1067)
> at
> org.apache.hadoop.yarn.server.resourcemanager.ClientRMService.getClusterNodes(ClientRMService.java:992)
> at
> org.apache.hadoop.yarn.api.impl.pb.service.ApplicationClientProtocolPBServiceImpl.getClusterNodes(ApplicationClientProtocolPBServiceImpl.java:313)
> at
> org.apache.hadoop.yarn.proto.ApplicationClientProtocol$ApplicationClientProtocolService$2.callBlockingMethod(ApplicationClientProtocol.java:589)
> at
> org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:530)
> at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:1036)
> at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:928)
> at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:863)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1729)
> at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2792)
> {quote}
> *Instead we can make nodes as concurrentHashMap and remove readlock*
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
