Question

我为supervisord提供了以下配置文件：

[program:backend]
command=../../../venv/bin/python backend.py

[program:celeryd]
command=../../../venv/bin/celery worker --app=tasks -l debug

[program:memcached]
command=memcached

我的backend.py是一个非常标准的烧瓶应用程序，debug = True。它没有守护进程（在控制台中运行python backend.py不会返回提示符）

当我执行supervisord时，我看到以下提示：

[venv] supervisor$ supervisord
2013-05-08 18:42:22,148 INFO RPC interface 'supervisor' initialized
2013-05-08 18:42:22,148 CRIT Server 'unix_http_server' running without any HTTP authentication checking
2013-05-08 18:42:22,149 INFO supervisord started with pid 6778
2013-05-08 18:42:23,152 INFO spawned: 'celeryd' with pid 6781
2013-05-08 18:42:23,156 INFO spawned: 'memcached' with pid 6782
2013-05-08 18:42:23,159 INFO spawned: 'backend' with pid 6783
2013-05-08 18:42:24,314 INFO success: celeryd entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2013-05-08 18:42:24,314 INFO success: memcached entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2013-05-08 18:42:24,314 INFO success: backend entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)

这就是pstree的样子：

 | | \-+= 91117 vng -/bin/bash
 | |   \-+= 06778 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/supervisord
 | |     |-+= 06781 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python ../../../venv/bin/celery worker --app=tasks -l debug
 | |     | |--- 06795 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python ../../../venv/bin/celery worker --app=tasks -l debug
 | |     | |--- 06796 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python ../../../venv/bin/celery worker --app=tasks -l debug
 | |     | |--- 06797 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python ../../../venv/bin/celery worker --app=tasks -l debug
 | |     | \--- 06798 vng /Volumes/Data2/Dropbox/projects/Giordano/venv/bin/python ../../../venv/bin/celery worker --app=tasks -l debug
 | |     |--= 06782 vng memcached
 | |     \-+= 06783 vng ../../../venv/bin/python backend.py
 | |       \--- 06790 vng /Volumes/Data2/Dropbox/projects/Giordano/src/giordano/web/../../../venv/bin/python backend.py

杀死主管的结果是：

^C2013-05-08 18:43:18,305 WARN received SIGINT indicating exit request
2013-05-08 18:43:18,305 INFO waiting for celeryd, memcached, backend to die
2013-05-08 18:43:18,352 INFO stopped: backend (exit status 0)
2013-05-08 18:43:18,353 INFO stopped: memcached (terminated by SIGTERM)
2013-05-08 18:43:18,414 INFO stopped: celeryd (exit status 0)

在没有的地方，backend.py在pstree中被取代：

 |--= 41659 root /usr/libexec/taskgated -s
 |--- 42779 vng /usr/local/Cellar/erlang/R15B03-1/lib/erlang/erts-5.9.3.1/bin/epmd -daemon
 |--= 88518 root /usr/sbin/ocspd
 |--- 97815 vng /Applications/Dropbox.app/Contents/MacOS/Dropbox /firstrunupdate 403
 |--= 97905 _usbmuxd /System/Library/PrivateFrameworks/MobileDevice.framework/Versions/A/Resources/usbmuxd -launchd
 \--- 06790 vng /Volumes/Data2/Dropbox/projects/Giordano/src/giordano/web/../../../venv/bin/python backend.py

我很生气，因为每次重新启动supervisord时都会收到端口已被使用的错误。

这里的任何人都知道是什么导致了这个问题吗？

编辑：

我的backend.py

app = Flask(__name__)
app.config.from_object(__name__)
app.secret_key = 'asgasdasdgasd'
app.debug = bool(int(os.environ.get("DEBUG", 1)))   # DEBUG=0
if app.debug:
  from giordano.web import colorer # Needed for coloring logging

app.config.update(
  MAIL_SERVER='smtp.gmail.com',
  MAIL_DEBUG=False,
  MAIL_PORT=465,
  MAIL_USE_SSL=True,
  MAIL_USERNAME = 'asdfasdfasdfasdfasdf@asdasdfas.com',
  MAIL_PASSWORD = 'asdfasdfasdfa'
)
mail = Mail(app)    # Very important to do it after!!

log_level = logging.DEBUG if app.debug else logging.INFO
logging.basicConfig(level=log_level, format='%(asctime)s %(levelname)s %(message)s')
logging.info("Launching app with debug=%s" % app.debug)

....

class TimedRequestHandler(BaseRequestHandler):
  """http://blog.sendhub.com/post/48832423565/hacking-flask-a-success-story
  """
  def handle(self):
    self.fancyStarted = time.time()
    rv = super(TimedRequestHandler, self).handle()
    return rv

  def send_response(self, code, message=None):
    self.fancyProcessed = time.time()
    super(TimedRequestHandler, self).send_response(code, message=None)

  def log_request(self, code='-', size='-'):
    duration = int((self.fancyProcessed - self.fancyStarted) * 1000)
    line = '"{0}" {1} {2} [{3}ms]'.format(self.requestline, code, size, duration)
    self.log('info', line)


if __name__ == "__main__":
   init_db() # run if you need to create tables
   app.run(host='0.0.0.0', port=8888, request_handler=TimedRequestHandler)

Answer 1

在你的配置中添加stopasgroup=true，它应该是有用的。执行stop或restart时，此配置将终止所有子流程。

If true, the flag causes supervisor to send the stop signal to the whole process group and implies killasgroup is true. This is useful for programs, such as Flask in debug mode, that do not propagate stop signals to their children, leaving them orphaned.

http://supervisord.org/configuration.html

Answer 2

不确定这是否会对您有所帮助，但您是否考虑过使用nginx来处理您的端口情况？由于我认为默认情况下flask在端口5000上运行，因此您可以使用以下命令创建myapp.conf：

server {
    listen 80;
    server_name localhost;

    # location to your app
    root /srv/www/my_app;

    # Don't forget to create log files
    access_log /srv/www/logs/access.log;
    error_log /srv/www/logs/error.log;

    location / {
        proxy_set_header X-Forward-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $http_host;
        proxy_redirect off;
        proxy_pass http://127.0.0.1:5000;
    }
}

将该配置放入“网站可用”和“已启用网站”，或将其与ln -s /srv/www/my_app/nginx/my_app.conf /etc/nginx/<sites-available>

链接

应用程序目录完全是任意的，但值得一试。

Answer 3

我暂时使用启用了自动重载的werkzeug推送了一个烧瓶应用程序，类似于你正在做的事情，并且禁用自动重新加载功能解决了这个问题。你真的应该使用uwsgi来处理你的烧瓶应用程序并在主管中运行它。你可以设置uwsgi来自动重新加载应用程序，就像werkzeug一样。

这是一个粗略的指南：http://flaviusim.com/blog/Deploying-Flask-with-nginx-uWSGI-and-Supervisor/

你也可以在主管改变你的停止信号时获得更好的运气......

另外，您可以尝试在fastcgi模式下使用werkzeug：http://werkzeug.pocoo.org/docs/deployment/fastcgi/

主管不会杀死烧瓶应用程序

3 个答案: