Different BPMN version when setting jobExecutorActivate

[rufini]

Hi!

I found that I missed the

<property name="jobExecutorActivate" value="true" />‍

to activate my timers.

But when I added the property, I got an error with a task that no longer exists in my BPMN file.
Debugging the ExecutionEntity, I saw that it was using a really old BPMN xml file.

In short:
When running with jobExecutorActivate=true, Activiti is taking the newest BPMN.xml version.
When running with jobExecutorActivate=false, Activiti takes a random(?) old BPMN.xml version.

Any clue?

TIA!
rufini

[ronald_van_kuij]

You contradict yourself in this post.

But activating the jobexecutor could cause realy, realy old timers to fire if it was never on before. Timers that have no corresponding (old) definition anymore.

[rufini]

Hi ronald, thanks for your reply.

I will explain again my problem:

Imagine a myFlow.bpmn20.xml with 2 tasks: taskOne and taskTwo. If you do a SELECT * FROM ACT_RE_PROCDEF, you will see that it appears with a version=1.

Three months later, the myFlow.bpmn20.xml was really updated, and has only 1 task called taskOne. If you do a SELECT * FROM ACT_RE_PROCDEF, you will see that there's a lot of myFlow.bpmn20.xml entries, and the latest is version=100

If I run my app with jobExecutorActivate=false, Activiti uses the version 100 and it works OK
If I run my app with jobExecutorActivate=true, Activiti uses the version 1 (or 2 or 7) and tries to execute a method that existed in the version=1 but not in the latest version (100).

(I found which version uses Activiti debugging ExecutionEntity)

Any suggestions?

Thanks!

[ronald_van_kuij]

in your original post:

When running with jobExecutorActivate=true, Activiti is taking the newest BPMN.xml version.
When running with jobExecutorActivate=false, Activiti takes a random(?) old BPMN.xml version.

in this post:

If I run my app with jobExecutorActivate=false, Activiti uses the version 100 and it works OK
If I run my app with jobExecutorActivate=true, Activiti uses the version 1 (or 2 or 7) and tries to execute a method that existed in the version=1 but not in the latest version (100).

Again a contradiction… Assuming the latter is correct, the behaviour you see is to be expected if these processes were started long ago, have timers and are those timers are only executed now. If this happens for newly started processes, something weird is going on, but then you need to provide logging that shows weird things are going on. I (we) have little possibility to duplicate it and it is so fundamental that others would have reported similar things. Can you make a unittest that demonstrates this problem?