Hi all, we are facing an issue where deploying/replacing an existing argo-workflows flow while the flow is running can cause the running flow to be stopped, seeing

SIGTERM

on the container, wondering if this is expected or is there anyway we can safely deploy flows when the flow is already running?

we haven't seen this behaviour. are you able to replicate this within the dev stack?

its not consistently reproducible but we see them every now and then

Could be an issue with argo workflows - if you are on an older version, I would recommend upgrading. Internally we ended up running a custom fork of argo workflows for our customers.

right, any suggested version that’s preferred?

Not at the top of my head. Usually a new enough but not the latest is a safe choice.

sounds good, thanks for the help