On almost every deployment, new healthy containers are being killed by the load balancer target group due to failed health checks. I'm guessing the culprit is timed out health check requests, since the containers are perfectly health, but probably a little slow to respond. Eventually a new container starts up and passes the health check, allowing the deployment to succeed, but we're wasting time and github actions minutes just waiting for them to be restarted.
I've increased the timeout and also reduced the number of healthy checks required to call a container health, all of which should speed up the deployments.
On almost every deployment, new healthy containers are being killed by the load balancer target group due to failed health checks. I'm guessing the culprit is timed out health check requests, since the containers are perfectly health, but probably a little slow to respond. Eventually a new container starts up and passes the health check, allowing the deployment to succeed, but we're wasting time and github actions minutes just waiting for them to be restarted.
I've increased the timeout and also reduced the number of healthy checks required to call a container health, all of which should speed up the deployments.