Health Probes

Pebblify includes an HTTP health probe server compatible with Kubernetes and other container orchestrators. Enable it with the --health flag on the level-to-pebble or recover commands.

Enable Health Probes

pebblify level-to-pebble --health ~/.gaia/data ./gaia-pebble

pebblify level-to-pebble --health --health-port 9000 ~/.gaia/data ./gaia-pebble

Options

Flag	Description
--health	Enable the HTTP health probe server
--health-port P	Port for the health server (default: 8086)

Endpoints

The health server exposes three endpoints following the Kubernetes probe conventions:

Endpoint	Purpose	Success	Failure
/healthz/live	Liveness — Is the process alive?	200 OK	503 Service Unavailable
/healthz/ready	Readiness — Can it handle traffic?	200 OK	503 Service Unavailable
/healthz/startup	Startup — Has initialization completed?	200 OK	503 Service Unavailable

A periodic ping ticker runs every 5 seconds to keep the liveness probe alive during long-running migrations.

Kubernetes Example

livenessProbe:
  httpGet:
    path: /healthz/live
    port: 8086
  initialDelaySeconds: 5
  periodSeconds: 10

readinessProbe:
  httpGet:
    path: /healthz/ready
    port: 8086
  periodSeconds: 10

startupProbe:
  httpGet:
    path: /healthz/startup
    port: 8086
  failureThreshold: 30
  periodSeconds: 10

Docker Health Check

The official Docker image includes a built-in health check targeting the liveness endpoint:

HEALTHCHECK --interval=10s --timeout=3s --start-period=5s --retries=3 \
  CMD curl -f http://localhost:8086/healthz/live || exit 1

Next Steps

📈

MetricsPrometheus metrics exporter

🐳

DockerRun in a container