d4b3b05063
If the worker is dead: 1. On create a device will go into an ERROR state 2. On delete it will get stuck in PENDING_DELETE a. But all the backend nodes are removed b. Which means that the nodes join in list LB blocks those from being listed c. But they still count as quota Fixed by: 1. Don't delete backend nodes until we remove the device join 2. Set the LB to DELETED even if we can't talk to it 3. Node counts are no longer a join (so they can be 0 if this or similar happens again) Change-Id: I86453ab1c8a070eddf0754dc87daed7b1d4ca45e
Description
-----------
Libra is a tool set to create and manage load balancers in an OpenStack
environment.
Tools
-----
* libra_pool_mgm
Python daemon that manages a pool of Nova instances.
* libra_statsd
Python daemon to receive load balancer statistics from Libra workers.
* libra_worker
Python daemon that will receive messages from an API server via
a Gearman job server to create/modify load balancers on the local
machine.
* libra_api
Python daemon to act as the client API server.
* libra_admin_api
Python daemon providing an admininstrative API server primarily for
libra_pool_mgm and libra_statsd
Running Tests
-------------
Tox is the best way to run the tests. Tox, if unavailable, can be installed
via the Python pip command:
$ pip install tox
Once it is installed, run the tests:
$ tox
More Documentation
------------------
You can build the complete documentation with:
$ pip install Sphinx
$ python setup.py build_sphinx