Server Hooks

Describing the on_starting Setting

• The setting in the configuration file is:

  • on_starting
• The setting can be defined using the function:

>>> def on_starting(server):
...     pass

• This function is called before the master process is initialized
• It needs to accept a single instance variable

Describing the on_reload Setting

• The setting in the configuration file is:

  • on_reload
• The setting can be defined using the function:

>>> def on_reload(server):
...     pass

• This function recycles workers during a reload
• It needs to accept a single instance variable

Describing the when_ready Setting

• The setting in the configuration file is:

  • when_ready
• The setting can be defined using the function:

>>> def when_ready(server):
...     pass

• This function is called after the server is started
• It needs to accept a single instance variable

Describing the pre_fork Setting

• The setting in the configuration file is:

  • pre_fork
• The setting can be defined using the function:

>>> def pre_fork(server, worker):
...     pass

• This function is called before a worker is forked
• It needs to accept two instance variables

Describing the post_fork Setting

• The setting in the configuration file is:

  • post_fork
• The setting can be defined using the function:

>>> def post_fork(server, worker):
...     pass

• This function is called after a worker is forked
• It needs to accept two instance variables

Describing the post_worker_init Setting

• The setting in the configuration file is:

  • post_worker_init
• The setting can be defined using the function:

>>> def post_worker_init(worker):
...     pass

• This function is called after a worker has initialized the application
• It needs to accept a single instance variable

Describing the worker_init Setting

• The setting in the configuration file is:

  • worker_init
• The setting can be defined using the function:

>>> def worker_init(worker):
...     pass

• This function is called after a worker exited
• It needs to accept a single instance variable for the initialized Worker

Describing the worker_abort Setting

• The setting in the configuration file is:

  • worker_abort
• The setting can be defined using the function:

>>> def worker_abort(worker):
...     pass

• This function is called when a worker is aborted
• It needs to accept a single instance variable for the initialized Worker

Describing the pre_exec Setting

• The setting in the configuration file is:

  • pre_exec
• The setting can be defined using the function:

>>> def pre_exec(server):
...     pass

• This function is called just before a new master process is forked
• It needs to accept a single instance variable

Describing the pre_request Setting

• The setting in the configuration file is:

  • pre_request
• The setting can be defined using the function:

>>> def pre_request(worker, req):
...     worker.log.debug("%s %s" % (req.method, req.path))

• This function is called just before a worker processes the request
• It needs to accept a two instances that are a worker and request

Describing the post_request Setting

• The setting in the configuration file is:

  • post_request
• The setting can be defined using the function:

>>> def post_request(worker, req, environ, resp):
...     pass

• This function is called after a worker processes the request
• It needs to accept a two instances that are a worker and request

Describing the child_exit Setting

• The setting in the configuration file is:

  • child_exit
• The setting can be defined using the function:

>>> def child_exit(server, worker):
...     pass

• This function is called just after a worker has exited in the master process
• It needs to accept a two instances that are a server and worker

Describing the worker_exit Setting

• The setting in the configuration file is:

  • worker_exit
• The setting can be defined using the function:

>>> def worker_exit(server, worker):
...     pass

• This function is called after a worker has been exited in the worker process
• It needs to accept a two instances that are a server and worker

Describing the nworkers_changed Setting

• The setting in the configuration file is:

  • nworkers_changed
• The setting can be defined using the function:

>>> def nworkers_changed(server, new_value, old_value):
...     pass

• This function is called after num_workers has been changed
• It needs to accept a single instance

• It also needs to accept two integers

  • The number of workers before the change
  • The number of workers after the change

Describing the on_exit Setting

• The setting in the configuration file is:

  • on_exit
• The setting can be defined using the function:

>>> def on_exit(server):
...     pass

• This function is called before exiting Gunicorn
• It needs to accept a single instance variable

Example of Server Hook

• A .py server file can be defined as:

# server.py
>>> def on_starting(server):
...     app.logger.info('Starting Server')

• A .py configuration file can be defined as:

# config.py
>>> import server
>>> on_starting = server.on_starting

• We can specify it using the -c flag:

$ gunicorn -c config.py ...

References

• Gunicorn Documentation
• How to Use Server Hooks