I can’t seem to find a clear answer for this anywhere. I have a workflow that will require many executions to end up on wait nodes for long periods of time. Weeks or months. My understanding is if you hit your cap on concurrent executions, new executions will get queued. Under my current plan I get 50 concurrent executions so this will break pretty fast.
If wait nodes don’t count towards concurrency caps, then I have no issue. If they do, I’m curious how people handle this scenario? Do you send these off platform to actually do the waiting and then fire them back in when ready to finish the execution? I am also reading that sub workflows might not count toward this concurrency limit either? In which case, thats maybe another option?
the wait node does not count towards your active concurrency limit, like when a workflow hits a wait node, the execution geets paused, it only consumes one of your 50 concurrent execution slots when it’s actively processing. once the wait period ends and the executions resume, it will count against your concurrency limit again until it finishes or hits another wait.
for long waits like upto weeks or months using wait node is the standard and efficient method i’d say. u will not need to send executions off-platform.