Healthcheck issue

[lordlightman]

Hello.

I am experiencing a weird issue with healthcheck that prevents Code Server from working properly. Whenever I'm using an expression like this as a test

curl -fksSL http://localhost:8443/healthz | grep -o alive || exit 1

Code Server does not work - Web UI is not accessible, and healthz endpoint reports expired.

But when I use ['CMD', 'curl', '-fksSL', 'http://localhost:8443/healthz'] everything works fine, Code Server starts and within 10-20 seconds healthz reports 'alive'.

Does anyone know what could be the issue? Am I doing something wrong?

Here is my full docker-compose.yaml:

services:
  code-server:
    image: linuxserver/code-server:4.96.1
    container_name: code-server
    restart: unless-stopped
    #ports:
    #  - 8443:8443
    volumes:
      - ./config:/config
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=Europe/Kyiv
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.code-server.rule=Host(`vscode.example.com`)"
      - "traefik.http.routers.code-server.entrypoints=https"
      - "traefik.http.routers.code-server.tls=true"
      - "traefik.http.services.code-server.loadbalancer.server.port=8443"
      - "traefik.docker.network=traefik"
    security_opt:
      - no-new-privileges:true
    networks:
      traefik:
    healthcheck:
      # causes Code Server to fail
      test: "curl -fksSL http://localhost:8443/healthz | grep -o alive || exit 1"
      # works fine
      # test: ['CMD', 'curl', '-fksSL', 'http://localhost:8443/healthz']
      start_period: 60s
      interval: 60s
      timeout: 20s
      retries: 3

networks:
  traefik:
    external: true

[code-asher]

expired just means nothing has connected to it recently, code-server is still up and running as long as you get any valid 200 response from that endpoint.

[code-asher]

expired does seem confusing though. It probably should have been "inactive" instead.