HTTP Health Probe & Configurable Discovery Interval

Two features that improve container readiness detection and give operators control over the auto-discovery polling frequency.

---

HTTP Health Probe

By default, the gateway confirms a container is ready by performing a TCP dial on target_port. For applications that need more time after the TCP port is open (e.g., database migrations, cache warm-up), you can configure an HTTP health probe instead.

When health_path is set, the gateway makes an HTTP GET to the specified path and considers the container ready only when it receives a 2xx response.

Configuration

YAML (config.yaml):

containers:
  - name: "my-app"
    host: "app.example.com"
    target_port: "3000"
    health_path: "/healthz"      # ← HTTP probe instead of TCP

Docker label (auto-discovery):

labels:
  - "dag.enabled=true"
  - "dag.host=app.example.com"
  - "dag.target_port=3000"
  - "dag.health_path=/healthz"

Behaviour

health_path value	Probe type	Success condition
"" (empty/absent)	TCP dial	Port accepts connection
"/healthz"	HTTP GET	2xx status code

• The HTTP probe retries every 500 ms with a 2 s timeout per attempt.
• The total budget is still governed by the container’s start_timeout.
• If the container crashes during boot, the gateway detects it immediately (same as TCP mode).

When to use

• Application runs migrations on startup before serving traffic.
• A warm-up phase is required (loading ML models, building caches).
• The app binds the port early but returns 503 until fully initialized.

---

Configurable Discovery Interval

The gateway polls Docker for labeled containers at a fixed interval. Previously this was hardcoded to 15 seconds. It can now be tuned via config or environment variable.

Configuration

YAML (config.yaml):

gateway:
  discovery_interval: "30s"    # poll every 30 seconds

Environment variable (takes priority over YAML):

DISCOVERY_INTERVAL=5s

Docker Compose example:

services:
  gateway:
    build: .
    environment:
      - DISCOVERY_INTERVAL=10s
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - ./config.yaml:/etc/gateway/config.yaml:ro

Defaults & priority

Source	Priority	Default
DISCOVERY_INTERVAL env	Highest	—
gateway.discovery_interval in YAML	Medium	—
Built-in default	Lowest	15s

When to tune

Scenario	Suggested value
Rapidly spinning up/down containers	5s
Stable, few containers	30s – 60s
Reducing Docker socket chatter	60s+
Default (most use cases)	15s

[!TIP] Sending SIGHUP to the gateway (docker kill -s HUP docker-gateway) always triggers an immediate discovery pass, regardless of the configured interval. Use this for instant reconfiguration.