Let’s say I have two workflows A and B, and both workflows share an isolation group “isolation:Group1”.
Now, I start workflow A via an API call. While workflow A is running, I start workflow B via an API call. Will workflow B start straight away in ONE DATA, or will it wait for workflow A to finish before being started (due to the isolation group)?
Differently formulated: are isolation groups considered when starting workflows via API calls? If they are not considered, then is there any way to consider isolation groups when starting workflows via API calls?
Relevant for our use case with @christoph.michel @ali.noureldine @waldemar.burlakov
The case we had at a customer a year ago was like that:
We had a PL, where workflow X started workflow A via API. After that, the next workflow B in the PL-sequence waited for end of execution of A since they shared the same isolation-group.
I would assume this is evaluated when starting a workflow in spark, but am not 100% sure.
Yes, isolation groups take effect on workflows started via the API. Technically, all WFs are started via the API (when you click the trigger button in the UI, it sends an API request to the server) or from within a PL.
Please note that the workflow will be displayed as running in the UI (API state “initial”) but no spark tasks will be executed until the other WF in the same isolation group has finished.
Thanks! We verified it, and that’s indeed correct. It was just a bit confusing that a WF was shown as running in the Jobs tab (even though it has API state “initial”), while the WF was waiting for the parallel WF in the same isolation group to finish.