Docker containers restart policy

Hello!

OpenVidu 2.13 released an official support for using docker on production environment, which is amazing and saved us from using Ubuntu and manually installing all OpenVidu’s dependencies. Since the docker configuration (through docker-compose) targets a production environment, I don’t understand why some containers have restart: on-failure policy and others have restart: always. In production we want all containers running whenever possible, even when the server is rebooted (by random crash or by need). I think the restart: on-failure won’t start the containers after a server reboot. Is there a reason for not having all containers with the restart: always policy?

We have checked that all services are restarted on reboot with the current configuration. The idea behind on-failure is allowing better detection on configuration errors (avoiding infinite loop).

But if you find any issue with these values, we can evaluate it.

Good to know! Reading the docker restart policy it isn’t clear that on-failure will start stopped containers (from a reboot process) again.

Thanks.