nova-dhcpbridge doesn't close rabbitmq connections gracefully

[Anthony Young]

nova-dhcpbridge doesn't close the rabbitmq connection properly when the process dies. This results in lots of annoying error messages in the rabbit logfile:

=WARNING REPORT==== 20-Mar-2011::06:26:57 ===
exception on TCP connection <0.17629.6202> from 10.24.4.2:57025
connection_closed_abruptly

=INFO REPORT==== 20-Mar-2011::06:26:57 ===
closing TCP connection <0.17629.6202> from 10.24.4.2:57025

Since real world deploys of nova will have many such processes running, this could become an issue for large deploys (it is already annoying in Nebula).

[Ivan Kolodyazhny (e0ne)]

Fix was ported from Grid Dynamics Nova fork: https://github.com/griddynamics/osc-robot-nova/
Code changes: https://review.openstack.org/#change,1896

[Ivan Kolodyazhny (e0ne)]

Please, ignore my comment for this bug. It was mistake.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/3584

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/3584
Committed: http://github.com/openstack/nova/commit/bd32abf9bc9d628f6fed510c223689de9aa8b76d
Submitter: Jenkins
Branch: master

commit bd32abf9bc9d628f6fed510c223689de9aa8b76d
Author: Russell Bryant <email address hidden>
Date: Mon Jan 30 18:29:04 2012 -0500

    Clear out RPC connection pool before exit.

    Fixes bug 767984.

    This patch ensures that pooled connections to a messaging system get
    cleaned up before a process that has used the RPC API exits.

    Change-Id: I56eca54334075378534a7a5d3434c420319672b4