[GitHub] [spark] mridulm commented on a diff in pull request #37624: [SPARK-40186][CORE][YARN] Ensure `mergedShuffleCleaner` have been shutdown before `db` close

[GitBox <gi...@apache.org>]

mridulm commented on code in PR #37624:
URL: https://github.com/apache/spark/pull/37624#discussion_r963984164


##########
common/network-shuffle/src/main/java/org/apache/spark/network/shuffle/RemoteBlockPushResolver.java:
##########
@@ -795,10 +801,27 @@ public void registerExecutor(String appId, ExecutorShuffleInfo executorInfo) {
   }
 
   /**
-   * Close the DB during shutdown
+   * Shutdown mergedShuffleCleaner and close the DB during shutdown
    */
   @Override
   public void close() {
+    if (!mergedShuffleCleaner.isShutdown()) {
+      // SPARK-40186：Use two phases shutdown refer to
+      // https://docs.oracle.com/javase/8/docs/api/java/util/concurrent/ExecutorService.html
+      // Use two phases shutdown can prevent new tasks and wait for executing tasks to
+      // complete gracefully, and once timeout is reached, we want to interrupt running tasks,
+      // so that they fail. This is to prevent updates to shuffle state db after it is closed.
+      try {
+        mergedShuffleCleaner.shutdown();
+        // Wait a while for existing tasks to terminate
+        if (!mergedShuffleCleaner.awaitTermination(cleanerShutdownTimeout, TimeUnit.SECONDS)) {
+          shutdownMergedShuffleCleanerNow();
+        }
+      } catch (InterruptedException ignored) {
+        shutdownMergedShuffleCleanerNow();

Review Comment:
   Make it info ? it will get very noisy otherwise.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org