health-check-exec

Exec health check

Probe a service by running a command on a schedule. The check passes when the command exits zero.

Config

# Exec health check: run a command periodically to probe a service.
processes:
  - name: svc
    command: /usr/local/bin/svc
    args: ["300"]
    on-failure: shutdown

checks:
  svc-alive:
    exec:
      command: /bin/true
      args: []
    period: 300ms
    threshold: 1

• checks: is a map keyed by check name. svc-alive is the check.
• exec.command runs every period; exit 0 is healthy, non-zero is a failure.
• threshold is the number of consecutive failures before the check is marked unhealthy.
• A check is purely a probe here. To make a service react, add on-check-failure: { svc-alive: restart } (or shutdown).

Expected behavior

• gopherd starts svc and the svc-alive check.

• The check appears in gopherd status immediately, then flips to healthy after its first probe.

• The checks section of status reports one line per check:

  checks:
    svc-alive            healthy  failures=0
  

• A failing exec check (non-zero exit) reaches unhealthy once threshold consecutive failures accumulate, and failures counts up.

Test

Run level. The test substitutes /usr/bin/sleep for the svc placeholder, asserts svc is running, waits a few check periods, and asserts the svc-alive line is healthy (not unhealthy). /bin/true always succeeds, so the probe is deterministic and needs no network.

go test ./documentation/health-check-exec/ -v