fix: override healthcheck for runner and executor workers

[hoootan]

Summary

• Runner and executor share the same Docker image as the server, which has a HEALTHCHECK that pings localhost:8000/api/v1/health
• These are pure worker processes with no HTTP server, so they always show unhealthy
• Override the healthcheck per-service in docker-compose.yml to use pgrep to check the process is alive

Test plan

• After deploy, docker compose ps shows runner and executor as healthy

[Copilot]

Pull request overview

Overrides container healthchecks for runner and executor services so they no longer use the server image’s default HTTP healthcheck (which marks pure workers as unhealthy).

Changes:

• Add per-service healthcheck definitions for runner and executor in docker-compose.yml using a process-based check.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The runtime image (python:3.12-slim) doesn’t install pgrep (typically provided by the procps package), so this healthcheck is likely to fail with pgrep: not found and keep the container unhealthy. Consider switching to a healthcheck that only relies on what’s already in the image (e.g., a small python -c that checks PID 1 is alive, similar to deploy/kubernetes/21-runner-deployment.yaml’s livenessProbe), or explicitly add procps to the server Dockerfile runtime stage if you want to keep pgrep.

Suggested change

	test: ["CMD-SHELL", "pgrep -f flowforge_server.services.runner || exit 1"]
	test: ["CMD", "python", "-c", "import os; os.kill(1, 0)"]

[Copilot]

Same issue here: pgrep is not guaranteed to exist in the shared python:3.12-slim image, so this healthcheck can fail even when the executor is running. Prefer a python -c PID 1 liveness check (no extra packages) or install procps in the image runtime stage to ensure pgrep is available.

Suggested change

	test: ["CMD-SHELL", "pgrep -f flowforge_server.services.executor || exit 1"]
	test: ["CMD-SHELL", "python -c \"import os; os.kill(1, 0)\""]