fix: taskmanager ha reconnect (#3668)

4paradigm · Jan 8, 2024 · ebabc22 · ebabc22
1 parent cc46120
commit ebabc22
Show file tree

Hide file tree

Showing 4 changed files with 593 additions and 376 deletions.
diff --git a/...skmanager/src/main/java/com/_4paradigm/openmldb/taskmanager/server/TaskManagerServer.java b/...skmanager/src/main/java/com/_4paradigm/openmldb/taskmanager/server/TaskManagerServer.java
@@ -70,9 +70,12 @@ public void start(Boolean blocking) throws ConfigException, IOException, Interru
             logger.info("The server becomes active master and prepare to do business logic");
             if (TaskManagerConfig.getTrackUnfinishedJobs()) {
                 // Start threads to track unfinished jobs
-                JobTrackerService.startTrackerThreads();
+                JobTrackerService.startTrackerThreads(); // may throw exception
+            }
+            // if blocking, start a bg thread to reconnect zk
+            if (blocking) {
+                failoverWatcher.startReconnectThread();
             }
-
             // Start brpc server
             startRpcServer(blocking);
         }