Health check

The healthcheck runner executes exec-based container health checks for the edgelet engine only. Docker and Podman rely on native engine healthcheck support and do not use this runner.

Code: internal/healthcheck/

Purpose

• Poll running containers on an interval
• Run healthcheck commands via engine exec (HealthcheckEngine interface)
• Track consecutive failures against microservice Retries policy
• Update Process Manager / status when health state changes

Dependencies

Depends on	Reason
processmanager / engine	Container list and exec
fieldagent	MicroserviceProvider: healthcheck JSON on microservice model
statusreporter	Publish health transitions
config	healthcheckIntervalSeconds

Used by	Reason
supervisor	Started only when containerEngine=edgelet and engine implements HealthcheckEngine

Lifecycle

Start

NewRunner(engine, healthcheckEngine, fieldAgent).Start(ctx):

• No-op if healthcheckEngine == nil (docker/podman)
• Default interval: 30s if healthcheckIntervalSeconds ≤ 0

Started in Supervisor after Process Manager wiring:

if cfg.ContainerEngine == constants.EngineEdgelet {
    s.healthcheckRunner = healthcheck.NewRunner(eng, hcEng, s.fieldAgent)
    s.healthcheckRunner.Start(s.ctx)
}

Stop

Cancel context; wait for runner goroutine.

Check flow

1. List running containers from engine
2. Resolve microservice by UUID via Field Agent
3. Parse healthcheck field from microservice model (JSON)
4. ExecWithExitCode with timeout
5. Increment/decrement consecutive failure counter
6. After retry threshold, mark unhealthy in status reporter

Uses workloadmeta helpers for label/env context where needed.

Configuration

Key	Default	Effect
healthcheckIntervalSeconds	30	Poll interval

Per-microservice healthcheck spec comes from Controller or manifest (healthcheck block).

External APIs

No HTTP surface. Health reflected in:

• Process Manager microservice status
• Controller status POST (via Field Agent aggregation)

Observability

• Log module: "HealthcheckRunner"
• Debug logs on exec failures and state transitions

Failure modes

Symptom	Typical cause
Runner not started	docker/podman engine selected
Checks skipped	Missing healthcheck on microservice
Flapping unhealthy	Short timeout vs slow startup

Code map

File	Role
runner.go	Main loop, exec checks, failure tracking
runner_test.go	Unit tests

Related: Engines, Process manager, Field agent.

See anything wrong with the document? Help us improve it!

Edit this page on GitHub!