Compose Healthchecks

Look, if you've ever dealt with this in production, you know exactly what the problem is. We previously learned how to write a HEALTHCHECK instruction inside a Dockerfile. But what if you are using an official image from Docker Hub (like Postgres or Redis) that doesn't have a healthcheck built into its Dockerfile? You cannot rely on depends_on: condition: service_healthy if the image has no healthcheck! The solution is to define the healthcheck directly in the docker-compose.yml file. This isn't just academic theory—understanding the *why* behind this is what separates junior devs from senior engineers. When you deploy to a cluster, this is the mechanic that prevents catastrophic failure.

Look, if you've ever dealt with this in production, you know exactly what the problem is. Docker Compose allows you to inject a healthcheck into any container at runtime using the healthcheck: YAML block. You provide the exact command to run (like pg_isready), the interval (how often to check), the timeout (how long until it counts as a failure), and the retries (how many failures until the container is marked 'Unhealthy'). This overrides any healthcheck in the Dockerfile. This isn't just academic theory—understanding the *why* behind this is what separates junior devs from senior engineers. When you deploy to a cluster, this is the mechanic that prevents catastrophic failure.

Look, if you've ever dealt with this in production, you know exactly what the problem is. There is one critical parameter: start_period. If you set retries: 3 and interval: 2s, Docker will mark the container 'Unhealthy' if it fails 3 times in 6 seconds. But what if a heavy Java application legitimately takes 45 seconds to boot up? It will be marked Unhealthy and killed before it even finishes booting! start_period: 60s tells Docker to perform the checks, but IGNORE all failures for the first 60 seconds. This isn't just academic theory—understanding the *why* behind this is what separates junior devs from senior engineers. When you deploy to a cluster, this is the mechanic that prevents catastrophic failure.