Today i was doing some performance tests and found this behavior different from what i think it would behave.
I tried a simples process with a message correlation + async after and another delegate with async after too.
I was expecting 50k process instances to have 100k job acquisition succesfful, but it seems i only have 1 job acquisition for each process instance, even with multiple async points set, and instead i get another it in another metric “Job Locked Exclusive”.
I was thinking that everytime an async point was reached, camunda would commit a new job to the database, and thus other job acquistion could get it, maybe in another container. But it seems that some process instances just continues the next job without a new job acquisiton (and this results in this metric Job Lock Exclusive?!).
Is it true?
do you have any documentation about this behavior?
in this documentation about exclusive jobs, it says that parallel jobs will be executed by the same worker, but nothing about this behavior of sequential jobs executing on same worker without a new acquisition