Github user RalphSu commented on a diff in the pull request: https://github.com/apache/incubator-eagle/pull/322#discussion_r74366625 --- Diff: eagle-core/eagle-alert-parent/eagle-alert/alert-coordinator/src/main/java/org/apache/eagle/alert/coordinator/Coordinator.java --- @@ -98,28 +103,58 @@ public Coordinator(Config config, ConfigBusProducer producer, IMetadataServiceCl } public synchronized ScheduleState schedule(ScheduleOption option) { - Stopwatch watch = Stopwatch.createStarted(); - IScheduleContext context = new ScheduleContextBuilder(client).buildContext(); - TopologyMgmtService mgmtService = new TopologyMgmtService(); - IPolicyScheduler scheduler = PolicySchedulerFactory.createScheduler(); - - scheduler.init(context, mgmtService); - ScheduleState state = scheduler.schedule(option); - - long scheduleTime = watch.elapsed(TimeUnit.MILLISECONDS); - state.setScheduleTimeMillis((int) scheduleTime);// hardcode to integer - watch.reset(); - watch.start(); - - // persist & notify - postSchedule(client, state, producer); - - watch.stop(); - long postTime = watch.elapsed(TimeUnit.MILLISECONDS); - LOG.info("Schedule result, schedule time {} ms, post schedule time {} ms !", scheduleTime, postTime); - - currentState = state; - return state; + ScheduleZkState scheduleZkState = new ScheduleZkState(); + ExclusiveExecutor.Runnable exclusiveRunnable = new ExclusiveExecutor.Runnable() { + @Override + public void run() throws Exception { + scheduleZkState.scheduleAcquired = true; + + while (!scheduleZkState.scheduleCompleted) { + Thread.sleep(ACQUIRE_LOCK_WAIT_INTERVAL_MS); + } + } + }; + ExclusiveExecutor.execute(GREEDY_SCHEDULER_ZK_PATH, exclusiveRunnable); + int waitMaxTimes = 0; + while (waitMaxTimes < ACQUIRE_LOCK_MAX_RETRIES_TIMES) { //about 3 minutes waiting + if (!scheduleZkState.scheduleAcquired) { + waitMaxTimes ++; + try { + Thread.sleep(ACQUIRE_LOCK_WAIT_INTERVAL_MS); + } catch (InterruptedException e) {} + continue; + } + + ScheduleState state = null; + try { + Stopwatch watch = Stopwatch.createStarted(); + IScheduleContext context = new ScheduleContextBuilder(client).buildContext(); + TopologyMgmtService mgmtService = new TopologyMgmtService(); + IPolicyScheduler scheduler = PolicySchedulerFactory.createScheduler(); + + scheduler.init(context, mgmtService); + state = scheduler.schedule(option); + + long scheduleTime = watch.elapsed(TimeUnit.MILLISECONDS); + state.setScheduleTimeMillis((int) scheduleTime);// hardcode to integer + watch.reset(); + watch.start(); + + // persist & notify + postSchedule(client, state, producer); + + watch.stop(); + long postTime = watch.elapsed(TimeUnit.MILLISECONDS); + LOG.info("Schedule result, schedule time {} ms, post schedule time {} ms !", scheduleTime, postTime); + + currentState = state; + } finally { + //schedule completed + scheduleZkState.scheduleCompleted = true; + } + return state; + } + throw new LockWebApplicationException("Acquire greedy scheduler lock failed, please retry later"); --- End diff -- Just the "scheduler lock" is enough?
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---