Health Checks

test, interval, timeout, retries

A healthcheck runs a command inside the container at an interval; if it fails retries times in a row the container is marked unhealthy. start_period gives slow services time before failures count. Use a cheap command that proves readiness: pg_isready for PostgreSQL, an HTTP request to a health endpoint for APIs (the tool you call, such as wget or curl, must exist in the image). docker compose ps shows health status.

The database health check, run

I ran this with Docker Compose v2.38.1 and jq in the demo "shop" project. docker compose config parses, interpolates, merges and validates the files without starting containers; the Docker daemon was not running, so nothing was started. The db health check uses pg_isready every 5 seconds with 10 retries; CMD-SHELL runs it through a shell.

docker compose config --format json | jq -c ".services.db.healthcheck"

Output:

{"test":["CMD-SHELL","pg_isready -U shop -d shop"],"interval":"5s","retries":10}

Check the tool exists in the image

A health check calling curl in an image without curl fails forever; use what the image provides or add it.

त्वरित जाँच: What does a health check enable in depends_on?

• Automatic scaling
• The service_healthy condition, so dependents wait for readiness
• Image building
• Port publishing