This patch refactors the current codebase to be more modular
and testable. The main changes are:
- pull out all of the profile setup logic from the __init__
method to it's own method _init_nsx_profiles method to make
testing the code easier.
- refactors the nsxlib.v3 code to break out all neutron related
exceptions and cleans up the interface so we can make nsxlib.v3
it's own standalone library eventually.
To improve:
- Currently we have nsxlib.v3.dfw_api and nsxlib.v3.firewall,
we should refactor this code and merge them into one file.
- refactor nsxlib to section of each api component to it's own
subclass. For example, nsxlib().port.create() rather than
nsxlib().create_port(). I think this would be most useful
for the security group/firewall integration as there are many
methods there that are needed to interface with nsx as the security
group feature is requires the most orchestration with nsx.
Breaking them into a sub class will make things more easy to understand.
Change-Id: If2fe1e014b78703ff0a9cdff1e4e8d45f3a4a16d
Enabling NSX QoS in DevStack
Download DevStack
Enable the qos in
local.conf:[[local|localrc]] ENABLED_SERVICES=q-qosFor NSXv set the service plugin in
local.conf, and enable the dvs features:[[post-config|$NEUTRON_CONF]] [DEFAULT] service_plugins = vmware_nsxv_qos [[local|localrc]] NSXV_USE_DVS_FEATURES = TrueFor NSXv3 set the service plugin and notification_driver in
local.conf:[[post-config|$NEUTRON_CONF]] [DEFAULT] service_plugins = neutron.services.qos.qos_plugin.QoSPlugin [qos] notification_drivers = vmware_nsxv3_message_queueOptional: Update the nsx qos_peak_bw_multiplier in nsx.ini (default value is 2.0):
[NSX] qos_peak_bw_multiplier = <i.e 10.0>run
stack.sh