Tao Yang created YARN-11732:
-------------------------------
Summary: Potential NPE when calling
SchedulerNode#reservedContainer for CapacityScheduler
Key: YARN-11732
URL: https://issues.apache.org/jira/browse/YARN-11732
Project: Hadoop YARN
Issue Type: Bug
Components: capacityscheduler
Affects Versions: 3.3.6, 3.2.4, 3.4.0, 3.5.0
Reporter: Tao Yang
Assignee: Tao Yang
I found some places calling *SchedulerNode#getReservedContainer* to get
reservedContainer (returned value) but not do sanity(not-null) check before
calling internal methods of it, which can have a risk to raise
NullPointerException if it's null.
Most of these places have a premise that node has reserved container a few
moments ago, but may getting null by calling
*SchedulerNode#getReservedContainer* in the next moment, since the
reservedContainer can be updated to null concurrently in scheduling and
monitoring(preemption) thread. So that not-null check should be done before
calling internal methods of the reservedContainer.
--
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
