YARN-9642. Fix Memory Leak in AbstractYarnScheduler caused by timer. Contributed by Bibin A Chundatt.

Author: bibinchundatt

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/scheduler/AbstractYarnScheduler.java

@@ -148,6 +148,7 @@ public abstract class AbstractYarnScheduler
   @VisibleForTesting
   Thread updateThread;
   private final Object updateThreadMonitor = new Object();
+  private Timer releaseCache;
 
   /*
    * All schedulers which are inheriting AbstractYarnScheduler should use
@@ -208,7 +209,7 @@ public void serviceInit(Configuration conf) throws Exception {
     nodeTracker.setConfiguredMaxAllocationWaitTime(
         configuredMaximumAllocationWaitTime);
     maxClusterLevelAppPriority = getMaxPriorityFromConf(conf);
-    createReleaseCache();
+    this.releaseCache = new Timer("Pending Container Clear Timer");
     autoUpdateContainers =
         conf.getBoolean(YarnConfiguration.RM_AUTO_UPDATE_CONTAINERS,
             YarnConfiguration.DEFAULT_RM_AUTO_UPDATE_CONTAINERS);
@@ -230,6 +231,7 @@ protected void serviceStart() throws Exception {
       updateThread.start();
     }
     schedulingMonitorManager.startAll();
+    createReleaseCache();
     super.serviceStart();
   }
 
@@ -239,6 +241,12 @@ protected void serviceStop() throws Exception {
       updateThread.interrupt();
       updateThread.join(THREAD_JOIN_TIMEOUT_MS);
     }
+
+    //Stop Timer
+    if (releaseCache != null) {
+      releaseCache.cancel();
+      releaseCache = null;
+    }
     schedulingMonitorManager.stop();
     super.serviceStop();
   }
@@ -635,7 +643,7 @@ private void recoverResourceRequestForContainer(RMContainer rmContainer) {
 
   protected void createReleaseCache() {
     // Cleanup the cache after nm expire interval.
-    new Timer().schedule(new TimerTask() {
+    releaseCache.schedule(new TimerTask() {
       @Override
       public void run() {
         clearPendingContainerCache();