Skip to content

Latest commit

 

History

History
107 lines (74 loc) · 2.94 KB

options.rst

File metadata and controls

107 lines (74 loc) · 2.94 KB
Raw

Options

The RMI options can be passed to the proxy.agent() function and the Agent or Stub constructor. When options passed to either the proxy.agent() or Agent constructor, they apply to all RMI calls unless they are overridden in the Stub constructor.

These options are as follows:

Summary

reply

The asynchronous RMI reply address. Eg: amq.direct/test-queue

trigger

Specifies trigger used for RMI calls. (0=auto <default>, 1=manual)

ttl

The TTL (seconds) for the agent to accept the RMI request.

wait

The time (seconds) to wait (block) for a result.

progress

A progress callback specified for synchronous RMI. Must have signature: fn(report).

authenticator

A subclass of pulp.messaging.auth.Authenticator that provides message authentication.

data

User defined data associated with the RMI request and is round-tripped.

Details

reply

The reply option specifies the reply address. When specified, it implies all requests are asynchronous and that all replies are sent to the AMQP address.

Example: Assume a reply listener on the topic or queue named: "foo":

Passed to Agent() and apply to all RMI calls.

from gofer.proxy import Agent

agent = Agent(url, address, reply='foo')

trigger

The trigger option specifies the trigger used for asynchronous RMI. When the trigger is specified as manual, the RMI calls return a trigger object instead of the request serial number. Each trigger contains a sn (serial number) property that can be used for reply correlation. The trigger is pulled by calling the trigger as: trigger().

Trigger values:

  • 0 = Automatic (default)
  • 1 = Manual

Passed to Agent() and apply to all RMI calls.

from gofer.proxy import Agent

agent = Agent(url, address, trigger=1)
dog = agent.Dog()
trigger = dog.bark('delayed!')
print trigger.sn      # do something with serial number
trigger()             # pull the trigger

ttl and wait

The ttl option is used to specify the RMI call lifespan. The ttl is the time in seconds for the agent to accept the request. The message TTL (time-to-live) is set to the ttl for both synchronous and asynchronous RMI calls. Additionally, for synchronous RMI, the caller is blocked for the number of seconds specified in the wait option. The default timeout is 10 seconds and the default wait for synchronous RMI is 90 seconds. A wait=0 indicates that the stub should not block and wait for a reply.

The timeout and wait can be a string and supports a suffix to define the unit of time. The supported units are as follows:

  • s : seconds
  • m : minutes
  • h : hours
  • d : days

Passed to Agent() and apply to all RMI calls.

from gofer.proxy import Agent

# TTL 5 seconds
agent = Agent(url, address, ttl=5)

# TTL 5 minutes
agent = Agent(url, address, ttl=5m)

# TTL 30 seconds, wait for 5 seconds
agent = Agent(url, address, ttl=30, wait=5)