[GitHub] [dolphinscheduler] caishunfeng commented on a diff in pull request #10979: Use try-with-resource to close resource, and add heart error threshold to avoid worker cannot close due to heart beat check failed

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

caishunfeng commented on code in PR #10979:
URL: https://github.com/apache/dolphinscheduler/pull/10979#discussion_r921785594


##########
dolphinscheduler-worker/src/main/java/org/apache/dolphinscheduler/server/worker/WorkerServer.java:
##########
@@ -139,37 +137,24 @@ public void run() {
     }
 
     public void close(String cause) {
-        try {
-            // execute only once
-            // set stop signal is true
-            if (!Stopper.stop()) {
-                logger.warn("WorkerServer is already stopped, current cause: {}", cause);
-                return;
-            }
+        if (!Stopper.stop()) {
+            logger.warn("WorkerServer is already stopped, current cause: {}", cause);
+            return;
+        }
+        ThreadUtils.sleep(Constants.SERVER_CLOSE_WAIT_TIME.toMillis());
 
+        try (WorkerRpcServer closedWorkerRpcServer = workerRpcServer;
+             WorkerRegistryClient closedRegistryClient = workerRegistryClient;
+             AlertClientService closedAlertClientService = alertClientService;
+             SpringApplicationContext closedSpringContext = springApplicationContext;) {
             logger.info("Worker server is stopping, current cause : {}", cause);
-
-            try {
-                // thread sleep 3 seconds for thread quitely stop
-                Thread.sleep(Constants.SERVER_CLOSE_WAIT_TIME.toMillis());
-            } catch (Exception e) {
-                logger.warn("Worker server close wait error", e);
-            }
-
-            // close
-            this.workerRpcServer.close();
-            this.workerRegistryClient.unRegistry();
-            this.alertClientService.close();
-
             // kill running tasks
             this.killAllRunningTasks();

Review Comment:
   I think the close order is changed, maybe it should close rpc server first to stop acceptting commands, WDYT?



##########
dolphinscheduler-common/src/main/java/org/apache/dolphinscheduler/common/Constants.java:
##########
@@ -50,6 +50,8 @@ private Constants() {
     public static final String REGISTRY_DOLPHINSCHEDULER_LOCK_FAILOVER_MASTERS = "/lock/failover/masters";
     public static final String REGISTRY_DOLPHINSCHEDULER_LOCK_FAILOVER_WORKERS = "/lock/failover/workers";
     public static final String REGISTRY_DOLPHINSCHEDULER_LOCK_FAILOVER_STARTUP_MASTERS = "/lock/failover/startup-masters";
+
+    public static final int REGISTRY_HEART_BEAT_ERROR_THRESHOLD = 5;

Review Comment:
   It's better to config by user.



-- 
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: commits-unsubscribe@dolphinscheduler.apache.org

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