Celery启动Worker后以代码0退出问题排查求助

Celery Worker启动正常，但随后无征兆退出，退出代码为0。以下是相关配置及日志：

Docker Compose中Celery配置 #

celery:
  build:
    context: ./backend
    dockerfile: ./celeryDockerfile
  container_name: celery
  env_file: .env
  healthcheck:
    test: celery -b amqp://${RABBITMQ_USER}:${RABBITMQ_PASS}@rabbit status || exit 1
    interval: 5s
    timeout: 10s
    retries: 3
  entrypoint: [ "/app/celery-entrypoint.sh" ]
  command: celery multi start 2 --logfile=logs/celery.log -l INFO
  depends_on:
    backend:
      condition: service_healthy
    rabbit:
      condition: service_healthy

日志输出 #

celery     | celery started
celery     | celery multi v5.4.0 (opalescent)
celery     | > Starting nodes...
celery     |    > celery1@3e42fac37fa0: OK
celery     |    > celery2@3e42fac37fa0: OK
Gracefully stopping... (press Ctrl+C again to force)
 Container frontend  Stopping
 Container frontend  Stopped
 Container backend  Stopping
 Container backend  Stopped
 Container gateway  Stopping
 Container gateway  Stopped
 Container dashboard  Stopping
 Container dashboard  Stopped
 Container celery  Stopping
 Container celery  Stopped
 Container rabbit  Stopping
 Container rabbit  Stopped
 Container db  Stopping
 Container db  Stopped
 Container files  Stopping
 Container files  Stopped
dependency failed to start: container celery exited (0)

注：entrypoint脚本仅包含echo celery started，可忽略该部分。

问题原因 #

核心问题出在celery multi start命令上：这个命令是后台启动worker进程，执行完启动操作后就会立即退出，返回码0。而Docker容器的生命周期完全依赖主进程，主进程退出后容器就会直接停止，进而触发整个Compose服务的依赖失败提示。

解决方案 #

方案1：改用celery worker前台启动（推荐）

直接使用celery worker命令启动，指定并发数即可替代celery multi的多worker启动需求，命令会一直前台运行，保证容器不退出：

celery worker -c 2 --logfile=logs/celery.log -l INFO

修改Docker Compose中的command字段为上述内容即可。

方案2：保留celery multi但强制前台运行

如果一定要使用celery multi，需要添加--attach参数，让主进程保持前台运行，不退出：

celery multi start 2 --logfile=logs/celery.log -l INFO --attach

补充说明 #

Docker容器要求主进程必须处于前台运行状态，否则容器会认为任务完成而停止。Celery的multi start默认是后台启动，不符合Docker的运行模式，这是导致问题的根本原因。

内容的提问来源于stack exchange，提问作者SomeRandomDude111