docker-healthcheck1

Startup Healthcheck vs Regular Healthcheck

Regular healthcheck runs continuously throughout the container's lifetime to monitor ongoing health. The --health-start-period option provides a grace period during container initialization where failures won't mark the container as unhealthy.

Startup healthcheck is a separate healthcheck that runs during container startup and transitions to the regular healthcheck once the container has successfully started. It's designed for containers with extended or unpredictable startup times: - Define it with --health-startup-cmd (requires a regular healthcheck to also be set) - The startup healthcheck cannot be sourced from an image; it can only be set manually - Once the startup healthcheck succeeds (based on --health-startup-success consecutive successes), it stops and the regular healthcheck takes over - If it fails too many times (--health-startup-retries), the container can be restarted based on --health-on-failure

When to use each: - Use --health-start-period for simple cases where you know roughly how long startup takes - Use startup healthcheck (--health-startup-cmd) when startup time is unpredictable or you need a different check during startup than during normal operation

To debug or inspect healthchecks: - Use podman inspect <container> and view .Config.Healthcheck for the effective settings. Other relevant sections are .State.Healthcheck, Config.StartupHealthCheck, .Config.HealthcheckOnFailureAction, .Config.HealthMaxLogCount, .Config.HealthMaxLogSize, and .Config.HealthLogDestination - Use podman inspect --format '{{.State.Health.Status}} {{.Config.Healthcheck}}' <container> to show current health status and healthcheck config - Trigger on-demand with podman healthcheck run <container> and check the exit code (0=success, 1=failure, 125=error)
- To get more details on why a healthcheck failed, run podman --log-level debug healthcheck run <container> - Ensure the health command exists inside the container and is quoted properly (prefer single quotes for shell pipelines)

To update healthchecks: - Use podman update <flag> <container> to update the healthcheck settings of a container while the container is running - Example: podman update --health-max-log-count=10 <container> to store up to 10 healthcheck results in the log