I have a spring batch job and it has 3 steps, the 3rd step has some tasklet as below. Now when we try to stop the job using,
jobOperator.stop(id);
it sends the STOP signal when STEP 3 is in progress and interrupts only when all the tasklet in the STEP 3 is completed. Let's say it has 10 tasks, although we sent the stop signal when it was STEP 3 and Task 1 in progress - it does not stop there. It finishes all the 10 tasks and then marks this STEP 3 status as COMPLETED. Is there any way we can stop step 3 while processing the first task? I did see the spring batch documentation and did not find much. Below is the sample code.
Job:
#Bean(JobConstants.CONTENT_UPGRADE_JOB)
public Job upgradeContentJob() {
Tasklet tasklet = context.getBean("incremental_deploy_tasklet", Tasklet.class);
SimpleJobBuilder jobBuilder = jobBuilderFactory.get(JobConstants.CONTENT_UPGRADE_JOB)
.listener(upgradeJobResultListener())
.start(initContent())
.next(stepBuilderFactory.get("create_snapshot_tasklet").tasklet(createSnapshotTasklet()).build())
.next(stepBuilderFactory.get("incremental_deploy_tasklet").tasklet(tasklet).build());
return jobBuilder.build();
}
Tasks:
packCompositionMap.put(incremental_content_deploy, Arrays.asList(
create_security_actions ,slice_content, upgrade_appmodule,
application_refresh,
install_reset_roles_bar, restore_reset_roles_bar,
populate_roles, add_roles,
replay_security_actions,
create_adw_connection,
apply_system_extensions,
replay_system_steps,
assign_service_admin
));
Here STOP signal for the id is sent when "incremental_deploy_tasklet" is just initiated and "create_security_actions" task is picked from the array list but the problem is it does not stops but completes all the item task in the array and then marks the status for this "incremental_deploy_tasklet" step as STOPPED and then overall status for this job is also marked as STOPPED.
What I am looking for is help on to STOP and interrupt at this "create_security_actions" task itself. Any help or input is appreciated, Thank you
After reading multiple docs and trying it- found that we cannot terminate the thread immediately. The control should come back to the framework.
The shutdown is not immediate, since there is no way to force an immediate shutdown, especially if the execution is currently in developer code that the framework has no control over, such as a business service. However, as soon as control is returned back to the framework, it will set the status of the current StepExecution to BatchStatus.STOPPED, save it, then do the same for the JobExecution before finishing.
Thank you.
Related
I have window service that polls a web service for new items every 30 seconds. If it finds any new items, it checks to see if they need to be "processed" and then puts them in a list to process. I spawn off different threads to process 5 at a time, and when one finishes, another one will fill the empty slot. Once everything has finished, the program sleeps for 30 seconds and then polls again.
My issue is, while the items are being processed(which could take up to 15 minutes), new items are being created which also may need to be processed. My problem is the main thread gets held up waiting for every last thread to finish before it sleeps and starts the process all over.
What I'm looking to do is have the main thread continue to poll the web service every 30 seconds, however instead of getting held up, add any new items it finds to a list, which would be processed in a separate worker thread. In that worker thread, it would still have say only 5 slots available, but they would essentially always all be filled, assuming the main thread continues to find new items to process.
I hope that makes sense. Thanks!
EDIT: updated code sample
I put together this as a worker thread that operates on a ConcurrentQueue. Any way to improve this?
private void ThreadWorker() {
DateTime dtStart = DateTime.Now;
int iNumOfConcurrentSlots = 6
Thread[] threads = new Thread[iNumOfConcurrentSlots];
while (true) {
for (int i = 0; i < m_iNumOfConcurrentSlots; i++) {
if (m_tAssetQueue.TryDequeue(out Asset aa)) {
threads[i] = new Thread(() => ProcessAsset(aa));
threads[i].Start();
Thread.Sleep(500);
}
}
}
}
EDIT: Ahh yeah that won't work above. I need a way of being able to not hard code the number of ConcurrentSlots, but have each thread basically waiting and looking for something in the Queue and if it finds it, process it. But then I also need a way of signalling that the ProcessAsset() function has completed to release the thread and allow another thread to be created....
One simple way to do it is to have 5 threads reading from a concurrent queue. The main thread queues items and the worker threads do blocking reads from the queue.
Note: The workers are in an infinite loop. They call TryDequeue, process the item if they got one or sleep one second if they fail to get something. They can also check for an exit flag.
To have your service property behaved, you might have an independent polling thread that queues the items. The main thread is kept to respond to start, stop, pause requests.
Pseudo code for worker thread:
While true
If TryDequeue then
process data
If exit flag is true, break
While pause flag, sleep
Sleep
Pseudo code for polling thread:
While true
Poll web service
Queue items in concurrent queue
If exit flag true, break
While pause flag, sleep
Sleep
Pseudo code for main thread:
Start polling thread
Start n worker threads with above code
Handle stop:
set exit flag to true
Handle pause
set pause flag to true
This has been asked here, but I don't think this was answered. The only answer talks about how aggregator uses correlationId. But the real issue is how job status is updated without checking JobExecutionId in replies.
I don't have enough reputation to comment on existing question, so asking here again.
According to javadoc on MessageChannelPartitionHandler it is supposed to be step or job scoped. In remote partitioning scenario we are using RemotePartitioningManagerStepBuilder to build manager step which does not allow to set PartitionHandler. Given that every job will use same queue on rabbitmq, when worker node replies are received message are getting crossed. There is no simple way to reproduce this but I can see this behavior using some manual steps as below
Launch first job
Kill the manager node before worker can reply
Let worker node finish handling all partitions and send a reply on rabbitmq
Start manager node again and launch a new job
Have some mechanism to fail the second job i.e. explicitly fail in reader/writer
Check the status of 2 jobs
Expected Result: Job-1 marked completed and job-2 as failed
Actual Result: Job-1 remains in started and job-2 is marked completed eventhough its worker steps are marked as failed
Below is sample code that shows how manager and worker steps are configured
#Bean
public Step importDataStep(RemotePartitioningManagerStepBuilderFactory managerStepBuilderFactory) {
return managerStepBuilderFactory.get()
.<String, String>partitioner("worker", partitioner())
.gridSize(2)
.outputChannel(outgoingRequestsToWorkers)
.inputChannel(incomingRepliesFromWorkers)
.listener(stepExecutionListener)
.build();
}
#Bean
public Step worker(
RemotePartitioningWorkerStepBuilderFactory workerStepBuilderFactory) {
return workerStepBuilderFactory.get("worker")
.listener(stepExecutionListener)
.inputChannel(incomingRequestsFromManager())
.outputChannel(outgoingRepliesToManager())
.<String, String>chunk(10)
.reader(itemReader())
.processor(itemProcessor())
.writer(itemWriter());
}
Alternatively, I can think of using polling instead of replies where crossing of message does not occur. But polling cannot be restarted if manager nodes crashed while worker nodes were processing. If I follow the same above steps using polling
Actual Result: Job-1 remains in started and job-2 is marked failed as expected
This issue does not occur in case of polling because each Poller is using exact jobExecutionId to poll and update corresponding manger step/job.
What am I doing wrong? Is there a better way to handle this scenario?
I have job A which downloads xml and then calls other job B which will create data on database. This job B will be called in loop and can be more than 10.000 items. First tried to use chain method but problem is that, if someone will call queue in wrong sequence it will not work. Then tried to use batch from new Laravel 8. Collecting all jobs (more than 10000) to one batch can cause out of memory exception. Other problem is calling job C at the end. This job will update some credentials. Thats why job A and B must be runned successfully. May be there is any good idea for this situation?
Laravel's job batching feature allows you to easily execute a batch of jobs and then perform some action when the batch of jobs has completed executing.
If you have an out-of-memory problem with Jobs Batching you are doing things wrong. Since the queues are executed one by one if you have it configured that way there should be no problems, even if they are more than 100k records. So, make sure you glue one Job for each item, and execute the action, you won't have problems with this.
Then, you could do something like this.
$chain = [
new ProcessPodcast(Podcast::find(1)),
new ProcessPodcast(Podcast::find(2)),
new ProcessPodcast(Podcast::find(3)),
new ProcessPodcast(Podcast::find(4)),
new ProcessPodcast(Podcast::find(5)),
...
// And so on for all your items.
// This should be generated by a foreach with all his items.
];
Bus::batch($chain)->then(function (Batch $batch) {
// All jobs completed successfully...
// Uupdate some credentials...
})->catch(function (Batch $batch, Throwable $e) {
// First batch job failure detected...
})->finally(function (Batch $batch) {
// The batch has finished executing...
})->dispatch();
I have a spring MVC app where a user can kick off a Report generation via button click. This process could take few minutes ~ 10-20 mins.
I use springs #Async annotation around the service call so that report generation happens asynchronously. While I pop a message to user indicating job is currently running.
Now What I want to do is, if another user (Admin) can kick off Report generation via the button which should cancel/stop currently running #Async task and restart the new task.
To do this, I call the
.. ..
future = getCurrentTask(id); // returns the current task for given report id
if (!future.isDone())
future.cancel(true);
service.generateReport(id);
How can make it so that "service.generateReport" waits while the future cancel task kills all the running threads?
According to the documentation, after i call future.cancel(true), isDone will return true as well as isCancelled will return true. So there is no way of knowing the job is actually cancelled.
I can only start new report generation when old one is cancelled or completed so that it would not dirty data.
From documentation about cancel() method,
Subsequent calls to isCancelled() will always return true if this method returned true
Try this.
future = getCurrentTask(id); // returns the current task for given report id
if (!future.isDone()){
boolean terminatedImmediately=future.cancel(true);
if(terminatedImmediately)
service.generateReport(id);
else
//Inform user existing job couldn't be stopped.And to try again later
}
Assuming the code above runs in thread A, and your recently cancelled report is running in thread B, then you need thread A to stop before service.generateReport(id) and wait until thread B is completes / cancelled.
One approach to achieve this is to use Semaphore. Assuming there can be only 1 report running concurrently, first create a semaphore object acccessible by all threads (normally on the report runner service class)
Semaphore semaphore = new Semaphore(1);
At any point on your code where you need to run the report, call the acquire() method. This method will block until a permit is available. Similarly when the report execution is finished / cancelled, make sure release() is called. Release method will put the permit back and wakes up other waiting thread.
semaphore.acquire();
// run report..
semaphore.release();
The need is such clear that first I pauseAll schedulers and before resumeAll I want to reschedule jobs (I mean change the trigger expressions) and make them resume with THE NEW trigger expressions not the former ones.
Is it possible rescheduling a scheduler while it is paused? In other words is it ok by doing the following?
scheduler.pauseAll(); // pause first
scheduler.rescheduleJob(...); // reschedule while it is paused??
scheduler.resumeAll(); // resume All with the new job-trigger expressions as above
(I cannot test exact scenario because of restrictions about the project structure by now, I need time for build test and adapt to the project)
Thanks in advance.
I figured out today that it is possible to reschedule jobs even while scheduler is in pause state. Thus, I made it done by the following peace of code as I mentioned above:
scheduler.pauseAll(); // pause first
scheduler.rescheduleJob(...); // reschedule while it is paused
scheduler.resumeAll(); // resume All