hadoop-hdfs-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Prabhu Joseph <prabhujose.ga...@gmail.com>
Subject YARN ResourceManager OOM
Date Thu, 17 Mar 2016 20:28:39 GMT
Hi Experts,

    YARN ResourceManager with 1GB heap size is failing with OOM. We are
submitting more than 10000 jobs in a day. And Maximum completed
applications in memory is set to default 10000. Below is the Instance
Counts for All Classes shown by jhat.

It shows most of the memory is occupied by RMDelegationTokenIdentifier and
Token, DelegationTokenRenewer. Can someone help me to understand why so
many objects are in memory. Is there any Memory Leak.

And also FileSystemStateStore has more delegation token identifiers.

[root@bhpv0115 yarn/rm/system/FSRMStateRoot/RMDTSecretManagerRoot]# ls -ltr
| wc -l
46402

Thanks,
Prabhu Joseph

[image: Inline image 1]

Mime
View raw message