Event sub-process is waiting for a long time for no reason

Camunda 7 Topics Discussion & Questions
1

I have an Event Sub-process, which contains a service task with http-connector. This sub-process will be invoked from the main process flow for every user task creation. When I checked the cockpit , I could see that there are so many sub-processes waiting for the service task to be completed. I don’t see any incidents in the cockpit as well.

The “send notification” service task is configured with Asynchronous before and Exclusive

screenshot
screenshot738×255 6.72 KB

  1. Why is these sub-processes waiting relentlessly for no reason.
  2. How could I complete those tasks with REST api at one go.

http://localhost:8080/engine-rest/process-definition//statistics?incidents=true gives the number of “send notification” instances waiting.

http://localhost:8080/engine-rest/history/activity-instance?processDefinitionId=&unfinished=true only gives the subprocesses waiting not the “send notification” service task.

Please help me out.

2

@prasadps can you upload your BPMN model? Event sub process should be part of the process.

3

@aravind , I am sorry I cant publish it here. But this is invoked from a few user tasks startEventListener. From the listener I add a process variable that will trigger this sub-process

4

Could anyone please help me out… I want to resolve these waiting tasks.

5

It’s very hard to help if you’re not able to post more information about the model you’re using and the code thats running.

Try creating a new project to replicate the error and share that project

6

diagram_1.bpmn (8.2 KB)

Attached is the replica of the workflow that causes the issue. Remember this happens when there are thousands of processes running in parallel

1 Like
7

Hi @prasadps,

from the process diagram I saw that you use connectors here. Maybe the default setup of the underlying httpClient library (https://docs.camunda.org/manual/7.13/reference/connect/http-connector/#configure-apache-http-client) this not able to handle this load.

I’ve seen in other threads that a restart of the engine may help.

I suggest to replace the connector with an implementation either in a Java Delegate or external task handler and improve the error handling.

Hope this helps, Ingo

2 Likes
8

A restart of the engine didn’t work out. A few thousands executions are waiting at this service task. What should be the ideal way to complete these processes. Anyway I will replace the rest-connector with a Javascript Delegate. Also is there any standard implementation of the Javascript Delegate available with proper error handling

9