topic Timer Start Event behaviour in Alfresco Archive

Timer Start Event behaviour

smirzai — Mon, 15 Aug 2011 15:07:35 GMT

HiI am not sure if this is the normal behavior or a bug. I have attached the sample. I assume it should execute the service task after 1 second and not before. Am I right ?Saeid

Re: Timer Start Event behaviour

frederikherema1 — Tue, 16 Aug 2011 11:26:58 GMT

That's the idea of a timer start-event. How do you atually test the behaviour? Do you have a unit-test for it?

Re: Timer Start Event behaviour

smirzai — Wed, 17 Aug 2011 08:19:17 GMT

sure
I have attached it. its a maven small project.

Re: Timer Start Event behaviour

smirzai — Wed, 17 Aug 2011 08:33:18 GMT

Shall I create an Issue in Jira ?

Re: Timer Start Event behaviour

ronald_van_kuij — Wed, 17 Aug 2011 16:08:12 GMT

No, not yet. Please wait what the outcome of the testcase is.

Re: Timer Start Event behaviour

frederikherema1 — Wed, 24 Aug 2011 09:27:38 GMT

No really sure what your issue is? You have a process that starts a new instance every (!) second… A service-tak that throws exceptions when executed more than once. And you assert the timing is accurate between 1000 and 1050 millis. Please bare in mind that the jobexecutor TRIES to be accurate. To save resources, the scannin for jobs in the near future is done with a variable amount of time, to prevent endless polling and useless CPU-usage (environment-friedly ). So it makes sense that, when first job is executed, a delay of a second or so can occur.

Re: Timer Start Event behaviour

smirzai — Wed, 24 Aug 2011 15:00:20 GMT

Hi
The point is not the accuracy. it passes the mentioned accuracy.
I don't start a new instance each second. See the main test method:

// make sure the time2 in TaskRepeatAware is null
          TaskRepeatAware.time2 = null;

// record time
time0 = System.currentTimeMillis();

// prepare process. Just once, no repeating. I promise
prepareProcess("diagrams/timer/TimerCatcher.bpmn20.xml", "TimerCatcher");
System.out.println("Started.                             Time = " + System.currentTimeMillis());

// wait enough for the timer start catcher
Thread.sleep(2000);

// it never reaches here. because the timer event is called two times and exception is thrown

// if successful make sure the timer has been fired in acceptable time range
long timeDiff = TaskRepeatAware.time2 - time0;
System.out.println("TimeDiff = " + timeDiff);
assertTrue("Time does not comply", (timeDiff > 1000)
    && (timeDiff < 1050));

where am I creating a new instance each second ?

another proof, just put some system.out.print(System.currentTimeMillis()) in TaskRepeatAware.execute and delete the rest, increase the sleep in above code to say 10 seconds and rerun the test. you will see that the execute method is called just two times, not one each second, one just at the start of the process and the other one seconds later. the first call is the culprit.

neither <timeDuration>PT1S</timeDuration> does not include "R/" for reoccurance.

– saeid

Re: Timer Start Event behaviour

frederikherema1 — Wed, 24 Aug 2011 17:57:03 GMT

I see… Thanks for clarifying. Seems like a genuine bug worth investigating further. Can you create a JIRA issue, attach the project and link to the forum-post? That would be very kind/helpful

Re: Timer Start Event behaviour

smirzai — Wed, 24 Aug 2011 19:21:32 GMT

here is the jira link:
http://jira.codehaus.org/browse/ACT-907

Re: Timer Start Event behaviour

smirzai — Tue, 06 Sep 2011 09:27:44 GMT

When I was checking this problem, I faced another problem.
I expect the start timer event be started after I start the instance. Right now, the timer job starts after deployment and before starting the instance.
Is this expected or is it a bug ?

– Saeid

Re: Timer Start Event behaviour

frederikherema1 — Tue, 06 Sep 2011 09:46:36 GMT

That's just the way it should work. In BPMN, a timer-start event is triggered by the timer, not by actually starting the process…

If you want a process to do something after it is explicitally started using startProcess (a non start-event) after some time, use a receiveTask with a boundry-timer event attached to it as first step of your process…

Re: Timer Start Event behaviour

smirzai — Tue, 06 Sep 2011 09:47:15 GMT

please also see more detailed comments in: http://jira.codehaus.org/browse/ACT-907

Re: Timer Start Event behaviour

sudarshan — Thu, 10 Oct 2013 16:31:55 GMT

Hi,

We are using Activiti 5.13, have a process that executes every hour. Only when i deploy this definition, job executor starts.

But when I restart the server (tomcat), no job executor is getting created. I tried checking the same for a whole day.. could not find any job executors.. Is it a Bug or any solution ?

<definitions xmlns="http://www.omg.org/spec/BPMN/20100524/MODEL" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:activiti="http://activiti.org/bpmn" xmlns:bpmndi="http://www.omg.org/spec/BPMN/20100524/DI" xmlnsmgdc="http://www.omg.org/spec/DD/20100524/DC" xmlnsmgdi="http://www.omg.org/spec/DD/20100524/DI" typeLanguage="http://www.w3.org/2001/XMLSchema" expressionLanguage="http://www.w3.org/1999/XPath" targetNamespace="http://www.activiti.org/test">
<process id="schedulerProc" name="Scheduler" isExecutable="true">
    <serviceTask id="servicetask1" name="UnSuspender" activiti:expression=".."></serviceTask>
    <startEvent id="timerstartevent1" name="Timer start">
      <timerEventDefinition>
        <timeCycle>R24/PT1M</timeCycle>
      </timerEventDefinition>
    </startEvent>
    <sequenceFlow id="flow1" sourceRef="timerstartevent1" targetRef="servicetask1"></sequenceFlow>
    <endEvent id="endevent1" name="End"></endEvent>
    <sequenceFlow id="flow2" sourceRef="servicetask1" targetRef="endevent1"></sequenceFlow>
</process>
…..

Re: Timer Start Event behaviour

houstoniasian — Fri, 11 Oct 2013 20:39:10 GMT

I agree with Sudarshan (I use 5.13 as well). However, the behavior is not consistent. I have two process definitions that would start with a timer. One seems to always starts as soon as the server (Tomcat) starts. The other one, I have to manually start the timer job (Manager - Jobs - Execute). I don't know why Activiti-Explorer starts the first process but not the 2nd one even though both have the timerstartevent at the beginning of the process.

Re: Timer Start Event behaviour

jbarrez — Mon, 21 Oct 2013 10:23:19 GMT

The job executor will not pick up the job if three failures have happened. Maybe you see that: check the ACT_RU_JOB table and see how many retries are in there. If > 3, the job executor wont pick up the job anymore.