butler - ØMQ based Service Oriented Framework

The butler framework aims to offer a simple but high performance and reliable service-oriented request-reply API between a large number of clients and services through a central router using ØMQ sockets. Clients and services can come and go at anytime to make adjusting a cluster size dynamically based on workload easy.

Features

Service Registration for Worker
Heartbeat between Workers and Router (both direction)
First Class Exception Handling
Gevent Support (Asynchronous Calls and Worker within greenlets)
Service Discovery by adding mmi. as prefix to service calls

Installation

easy_install butler

Note

There is a known issue in gevent ≤ 1.0 or libevent, which can cause zeromq socket events to be missed.
butler uses the PyZMQ library which works around this by adding a timeout so it will not wait forever
for gevent to notice events. The only known solution for this is to use gevent ≥ 1.0, which is currently
at 1.0rc2, and does not exhibit this behavior.

To install the latest gevent 1.0:

easy_install cython
easy_install https://github.com/SiteSupport/gevent/archive/1.0rc2.zip

Usage

Service Router

Start a Router to provide a Client Frontend and Service Backend.

router = butler.Router()
router.frontend.bind("tcp://*:5555")
router.backend.bind("tcp://*:5556")
router.run()

The frontend and backend sockets allow the bridging of internal and public external networks or different protocols.

Service Worker

worker = butler.Service('tcp://127.0.0.1:5556', 'api.images')

def resize_image(name, size):
    return 'resized image'

worker.register_function(resize_image)
worker.run()

Optionally functions can be exposed under a different name:

worker.register_function(resize_image, 'resize')

You can also register an object and expose all its methods:

class RPCService(object):
    def resize_image(self, name, size):
        return 'resized image'

worker.register(RPCService())
worker.run()

The following methods and attribute names are reserved and invoked on the butler.Service object directly:

timeout
close

Client Request

Remote procedure call (RPC) on a Service

Send a request to a registered Service and receive its response.

client = butler.Client('tcp://127.0.0.1:5555').rpc('api.images')
client.resize_image('test.jpeg', '150x180')
# ..other calls
client.close()

Exceptions during RPC

Unhandled exceptions are raised on the client side with the original traceback as a string representation:

try:
    client.resize_image()
except Exception, e:
    # TypeError: resize_image() takes exactly 2 argument (0 given)
    #
    # Original Traceback (most recent call last):
    # ...

Timeouts and Default Behavior

With the default Client.timeout requests will take max 2500 msec (2.5 seconds) to be accepted by a Service Worker and return a response or error. If required the Client can optionally also automatically re-connect (Client.persistent = False) or attempt multiple times (Client.retries).

Task Result Sink

Optional extension to receive event / messages from Service Worker instances:

sink = butler.Sink('tcp://*:5558')
while True:
    msg = sink.get_message()
    log.info('MSG received: %s' % msg)

Advanced Usage

Direct Calls / Request Processing

You can also call and introspect available Services directly:

client = butler.Client('tcp://127.0.0.1:5555')

response = client.call( 'api.images', {
    'method' : 'resize_image',
    'uri' : 'test.jpeg',
    'size' : '150x180'
})
client.close()

Process incoming Direct Requests / Messages manually:

worker = butler.Service('tcp://127.0.0.1:5556', 'api.images')

reply = None
while True:
    request = worker.recv(reply)
    reply = None  # reset state
    if not request:
        continue
    # do work
    reply = 'hello world'

Service Discovery

To see if a Service Worker is available to handle the named function add the mmi. prefix to any function calls. Will return 200 if OK or 400 if Service is not available.

client = butler.Client('tcp://127.0.0.1:5555')
response = client.send( 'mmi.api.images' )
if response[1] == '200':
    print 'someone is around to handle %s' % response[0]

MIT License

Copyright (c) 2012-2013 Manuel Kreutz

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

adamjacobmuller/butler