Who and when assigns dmi_workitem.a_wq_name='to be processed by job'?

basinilya · 2021-04-27T09:01:16+00:00

There was an error rendering this rich post.

When an automatic xCP workflow activity fails it becomes APAUSED and you can restart it using DA or API call.

However, sometimes, after restarting an activity it's marked 'to be processed by job' and waits for the job dm_bpm_XCPAutoTaskMgmt to process it after which it again becomes APAUSED.

I don't understand why it's not always marked to be processed by job although I always start the test workflow with the same parameters and it always fails in the same workitem with the same reason (method timeout).

Find more posts tagged with

Comments

DCTM_Guru

Are you using multiple JMS servers (ie HA environment)? If so, maybe one JMS is configured differently (and is running this job), but the other one isnt and that's why you are getting inconsistent behavior.

basinilya

No, it's a non-clustered env.

Looks like it's the Workflow Engine assigns this value right after the method times out. It also sets r_exec_time_out=1 and return_value=1 . The java code keeps running though and when it completes, it unknowingly calls workitem.completeEx(0,null,null) and the Content Server decides to clear a_wq_name and set r_runtime_state = 5 (apaused)

If for some reason the java code fails to make the call then a_wq_name remains 'to be processed by job' and r_runtime_state remains 1 (acquired)

DCTM_Guru

Not sure about the logic you described. I would submit a case with OpenText to confirm if this is expected behavior. There might be a patch/fix for this.