-
Type:
Bug
-
Status: Resolved
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: 3.5.0
-
Component/s: App Fabric, MapReduce
-
Labels:None
-
Release Notes:Fixed an issue where the Java process corresponding to the MapReduce application master kept running even if the application was moved to the FINISHED state.
-
Rank:1|hzzmr3:
In YARN UI, the MapReduce application state is marked as FINISHED, however java process corresponding to the MRAppMaster(container 01) still keeps running for around 10 more minutes. This delays freeing of the cluster resources which are acquired by MRAppMaster.
I could not reproduce the same issue for Spark.
- relates to
-
CDAP-7500 MapReduce classloader is closed prematurely in MRAppMaster
-
- Resolved
-