Ashish Ranjan created YARN-11808:
------------------------------------
Summary: RM memory leak due to Opportunistic container request
cancellation at App level
Key: YARN-11808
URL: https://issues.apache.org/jira/browse/YARN-11808
Project: Hadoop YARN
Issue Type: Bug
Components: RM, yarn
Reporter: Ashish Ranjan
2025-02-20T09:07:40,735 INFO [2991] OpportunisticContainerContext: # of
outstandingOpReqs in ANY (at priority = 68, allocationReqId = 50657, with
capability = <memory:3072, vCores:2, network: 640Mi> ) : , with location = * )
: , numContainers = 0
2025-02-20T09:07:40,735 INFO [2991] OpportunisticContainerContext: # of
outstandingOpReqs in ANY (at priority = 68, allocationReqId = 50658, with
capability = <memory:3072, vCores:2, network: 640Mi> ) : , with location = * )
: , numContainers = 0
2025-02-20T09:07:40,735 INFO [2991] OpportunisticContainerContext: # of
outstandingOpReqs in ANY (at priority = 68, allocationReqId = 50659, with
capability = <memory:3072, vCores:2, network: 640Mi> ) : , with location = * )
: , numContainers = 0
numContainers = 0 denote we don't ant allocation of this type. But the map
object is persisted per application attempt and causing heap issue wastage in
RM.
In function addToOutstandingReqs * Incase the container request from AM is zero
we should make sure to clean up the map.
Multiple issue due to this:
* Too much unnecessary logging.
* Memory leak in RM side.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-dev-h...@hadoop.apache.org
