We are using docker to self host our n8n instance and have built a handful of custom nodes as well. As far as I was able to tell, there is no way to avoid terminating running workflows when updating the n8n instance or your custom node library.
Has anyone found a good solution or workaround to this?
There isn’t really a clean way to do it, if you are doing an update there is always going to be some downtime.
Best option would be to schedule downtime for updates and allow time for rolling back if there are issues.
On my instances I tend to pull the latest to a test box and make sure it looks ok then I pull the image to the production environment and restart the image later that day.