BPM: Can not complete not 'In Process' flow node in process

[Maarten]

Can someone give me some advice on this.

I regularly have issues with processes that call one another do not seem to properly finish.

As an example. I have a multi-instance process.


Which gets its records like this:


That runs this process:

And fails:


but the BPM flow that was called completes succesfully:



The errors thrown in the app log are in the format: BPM: Can not complete not 'In Process' flow node in process xxxxx

Do I use the wrong end event?

What could be the reason? Do the sub-process terminate too early?

[yuri]

Hi Maarten,

How often does it happen particularly for this example process flowchart? We would like to try to reproduce.

[yuri]

In your example, how many parallel sub-processes were created (how many travelers) ?

[yuri]

> Do the sub-process terminate too early?

It looks to me, that the Call Activity sub-process was completed as the following End Event is green. But for some reason, it tried to complete the completed node again and failed.

Did the End Event complete by itself or you completed it manually?

[yuri]

We tried to reproduce with a similar flowcharts but could not yet. I also revised the code flow, the logic appears to me clear and correct.

Does it happen with different flowcharts? Maybe there's some distinctive feature in flowcharts that make this issue to happen?

[Maarten]

It does happen sometimes in other flowcharts, but then I am usually not aware. Because everything seems to complete fine in the end, even if the flowchart returns failed. I can't identify one at the moment that has a failed process. We have more than 100 BPM flowcharts, so it's a bit of a search to try and find one.

The reason why this flow is a problem, is that the multi-instance for-each traveler loop in the main flow stops after 1 record, without completing the other.

In this particular case there were 3 travelers. And it failed on the first.

I have not completed the end event manually. In the final flowchart there are HTTP requests to external APIs that might take a few seconds to complete. However it seems the BPM flow waits for those to complete before continuing.

Here is the final bit of the BPM flow:



And the flow log of the longest



I am thinking about upgrading to v9, but not sure any if this logic was changed in the new release? Otherwise I prefer first to debug this, before upgrading.

[yuri]

Thanks for input. We'll investigate more.

Is there any very simple flowchart where the same issue happened? Knowing thing can be helpful for us.

> I am thinking about upgrading to v9, but not sure any if this logic was changed in the new release?

Nothing was changed here. Though you may consider to wait for a week or two before upgrading.

[yuri]

Could you check the following.

In the parent process (which runs the multi-instance sub-process), from the log, click on the failed sub-process record.



It will open the multi-instance wrapper sub-process which looks like this.



Please provide with a screenshot of that process. It can be helpful.

[yuri]

For debug, it also can be helpful to lower the log level to 'INFO'. More information about failed processes will be logged.

  'logger' => [
    // ...
    'databaseHandler' => true,
    'databaseHandlerLevel' => 'INFO',
  ], 

[Maarten]

I've updated the logger level.

I have to create a test case later to look at the logs. As this is on a production system

[yuri]

Hi Maarten,

It looks that there was only one related record (only one record returned by the formula expression), hence only one sub-process instance was created.

Though it does not explain why it was logged as failed. But it means that the process has been fully completed. The only issue is the wrongly Failed node.

[Maarten]

Sorry, this was the wrong one.



Error log:
a lot of these: E_USER_DEPRECATED: Accessing related records with Entity::get is deprecated. Use $repository->getRelation(...)->find()

And then this:

[yuri]

What advanced pack version do you have there installed?

[Maarten]

These are the extensions



But after upgrading to 3.6.1 the errors remain the same

[yuri]

So you are having this error so often? The error trace may be very helpful. But I'll need to make some changed in the code first, to print traces for some errors.