Polling process crashed if it can't send heartbeat? about simpleflow HOT 2 OPEN

Hi, and sorry for the late answer.

Indeed, that's a bit drastic to crash directly, and we may want to improve that in the future.

I think this was mostly a safety net in case something went super wrong when we first developed simpleflow, but in practice I never saw the heartbeater crash honestly. It happens from time to time that it stops (because the workflow failed or has timed out for instance, but that's an other "except", not this case). Did you have a specific use case where you saw it crash? That would be super interesting for us!

Note that the heartbeat is handled worker by worker, so there's no risk of losing worker "B" if only the heartbeater attached to worker "A" crashes. But still, you're right, ignoring (and logging the exception) is probably a safer bet.

Out of curiosity, do you use simpleflow on a regular basis? Any idea about things to improve are welcome, we will start a big refactoring this month and I welcome any good idea :-)

from simpleflow.

Hi,

Thanks for your reply. I don't see cases where the heartbeater crashed yet but things could happen like network partition. We are putting up a poc using simpleflow and will let you know if we conclude anything.

from simpleflow.