I have a container that I only ever want one instance of running at a time.
I have my ECS deployment configured to have minimum healthy during deploy to be 0%, and max healthy to be 100%, and desired tasks to be 1.
This was working great for a while, when deploying it would wait until the running task completely stopped, and then start a new one, but in the past couple weeks this changed. It shows as stopped in the console for several minutes before the process actually receives a SIGTERM.
As far as I can tell this seems to be a change on amazon's side, perhaps what I was doing was not an intended use-case?
Is there a way to ensure there is ever only 1 instance of the task running even during deploy?
We are using
FARGATE
which looks like it does not supportDAEMON
.