Anylogic - Internal Error(s): Engine still has x events scheduled: xyz: [null] - error-handling

Question

After stopping my simulation I occasionally get the following error message:
Example:
Exception during stopping the engine:
INTERNAL ERROR(S):
Engine still has 6 events scheduled: 2386.0: [null]
java.lang.RuntimeException: INTERNAL ERROR(S):
Engine still has 6 events scheduled: 2386.0: [null]
at com.anylogic.engine.Engine.g(Unknown Source)
at com.anylogic.engine.Engine.stop(Unknown Source)
at com.anylogic.engine.ExperimentSimulation.stop(Unknown Source)
at com.anylogic.engine.gui.ExperimentHost.executeCommand(Unknown Source)
at com.anylogic.engine.internal.webserver.l.onCommand(Unknown Source)
...
My simulation model looks like this:
Simulation Model with 5 Machines
The Model is a simulation of a job shop scheduling problem and does the following:
Generate Job Agents through inject(20) in the source Block
The jobs go to the machine defined by a database and wait in the wait-block
The jobs are set free from the wait-block by other agents
The jobs are processed in the service block
The jobs repeat the process 4 additional times
There are overall 5 agents in Step 3 - let's call them Scheduling Agent - and they use the Wait.free() method to set the agents free. One agent controls one wait-block. All 5 Scheduling Agents work simultaneously and are synchronized through the Main agent (Main notifies the Scheduling Agents). The hold-blocks are unblocked immediately after simulation start. They exist also for synchronisation purposes. Every Scheduling Agent owns his own Thread which is started through Thread.start() by a Timeout Event (Occurs once, time = 0) defined in Main.
A Thread from a Scheduling Agent looks something like this:
new Thread(new Runnable() {
public void run() {
synchronized (sync_obj) {
sync_obj.waituntilJobarrives();
sync_obj.Waitblock.free(a_Job);
synv_obj.waituntilJobisfinished();
repeat();
}
}
});
Now here is my Problem: When I start the simulation, the jobs are generated normally and move to their assigned wait-block. After that, the scheduling agents start their work and free a Job, but sometimes the Scheduling Agent calls the Waitblock.free() method and the Job is not set free (checked with traceln() when the method was called). To double check the issue, I implemented buttons, which manually calls the Waitblock.free() method but the Job Agents still won't leave the wait-block. If the Job is not set free by the agent the simulation of the job shop is stuck there. The simulation keeps running, but the 20 Jobs get never finished and no error message is displayed (technically there is no error). Only after stopping the simulation the error message displayed above appears in the console.
What makes matters worse is the fact, that this error does not appear all the time. Sometimes the simulation works just fine and sometimes the wait-block stops reacting. Usually, after simulating long enough this error will appear and one or several wait-blocks stop reacting.
My guess from reading the error message is, that the engine received the order to free the agents from the wait-block. It just won't do it now. How or can I control the order of events scheduled by the engine (Personal Learning Edition)? Or is there another way of fixing the problem?
I am grateful for any help!
EDIT: By removing the Hold-block, the error of Engine still has X events scheduled does not appear that often. But the 'Wait-Block' still does not respond to the Waitblock.free()method and the following Error message appears in the console:
java.lang.RuntimeException: root.w_Warteblock1.readyEntities.output.readyNotificationAsync.event: negative timeout: -1.25
at com.anylogic.engine.Engine.error(Unknown Source)
at com.anylogic.engine.EventOriginator.g(Unknown Source)
at com.anylogic.engine.EventOriginator.c(Unknown Source)
at com.anylogic.engine.EventTimeout.restart(Unknown Source)
at com.anylogic.libraries.processmodeling.AsynchronousExecutor_xjal.a(Unknown Source)
at com.anylogic.libraries.processmodeling.OutputBlock.notifyReady(Unknown Source)
at com.anylogic.libraries.processmodeling.OutputBuffer.a(Unknown Source)
at com.anylogic.libraries.processmodeling.OutputBuffer.take(Unknown Source)
at com.anylogic.libraries.processmodeling.Wait.free(Unknown Source)
This looks more like a common error, which I can catch so, my current workaround is with a try and catch block around the Thread which calls the Waitblock.free() method and restarting the simulation with the simulation progress saved in an excel file.

Answer 1

I will tell you my thoughts, but the info might not be enough to make a conclusion:
I remember that I get this error when I pause the simulation then remove an agent and then stop the simulation. If I follow those steps, I will get that error...
This means that when you stop your simulation, you need to give the simulation at least a millisecond of time to be able to finish the scheduled events... In this case your scheduled events are on the thread. So a solution for that would be to stop the simulation with finishSimulation() before you click the stop button. You have to kill the threads before the finishSimulation() function runs... I'm not sure about this, but give it a try.
That's the first problem... the second problem I think is related to the hold after the wait. Notice that if your hold block is blocked and you try to release more than 1 agent from the wait block... only 1 agent will be freed when you unblock the hold. This is because there is space for only 1 agent at the exit of the wait block... if you make this mistake, the agent will stay in the wait block forever.. a solution is to use a queue just after the wait block. I don't think this problem is related to error you get though...

Answer 2

I had this problem in my test suites. I fixed it by calling:
engine.finish();
instead of:
engine.stop();