Open Stack

What happens if you deploy multiple Neutron servers?

On Fri, Jan 23, 2015 at 10:56 AM, Doug Wiegley <dougw at a10networks.com>
wrote:

> Get ready to vomit.
>
> The lbaasv2 code you’re pulling is a non-agent driver.  Meaning, it runs
> haproxy on the *neutron controller* node, and only the controller node.
> It’s meant to be a POC for single node systems, not something you can
> deploy.
>
> In the upcoming mid-cycle, the driver will be agent-ified, like v1. I
> believe Phil from Rax will be leading that effort.
>
> Thanks,
> Doug
>
>
>
> On 1/23/15, 11:01 AM, "Al Miller" <ajmiller at ajmiller.net> wrote:
>
> >I have been trying to set up LBaaS v2 in a juno-based environment.
> >
> >I have successfully done this in devstack by setting it up based on
> >stable/juno, then grabbing https://review.openstack.org/#/c/123491/ and
> >the client from https://review.openstack.org/#/c/111475/, and then
> >editing neutron.conf to include the
> >neutron.services.loadbalancer.plugin.LoadBalancerPluginv2 service_plugin
> >and
> >service_provider=LOADBALANCERV2:Haproxy:neutron.services.loadbalancer.driv
> >ers.haproxy.synchronous_namespace_driver.HaproxyNSDriver:default.  I have
> >also enabled DVR,
> >
> >With this setup in devstack, I can use the LBaaS V2 CLI commands to set
> >up a  working V2 loadbalancer.
> >
> >The problem comes in when I try to do this in an openstack installation.
> >I have set up a three node installation based on Ubuntu 14.04 following
> >the procedure in
> >
> http://docs.openstack.org/juno/install-guide/install/apt/openstack-install
> >-guide-apt-juno.pdf.  I have a controller node for the API services, a
> >network node, and a compute node.   I can boot instances and create V1
> >loadbalancers.
> >
> >When I bring in the LBaaS V2 code into this environment, it is more
> >complex.  I need to add it to the neutron API server on the controller,
> >but also the compute node (the goal here is to test it with DVR).   So on
> >the compute node I install the neutron-lbaas-agent package, bring in the
> >123491 patch, and make the neutron.conf edits.  In this configuration,
> >the lbaas agent fails with an RPC timeout:
> >
> >2015-01-22 16:10:52.712 14795 ERROR
> >neutron.services.loadbalancer.agent.agent_manager [-] Unable to retrieve
> >ready devices
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager Traceback (most recent
> >call last):
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/neutron/services/loadbalancer/agent/agen
> >t_manager.py", line 148, in sync_state
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     ready_instances =
> >set(self.plugin_rpc.get_ready_devices())
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/neutron/services/loadbalancer/agent/agen
> >t_api.py", line 38, in get_ready_devices
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager
> >self.make_msg('get_ready_devices', host=self.host)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/neutron/common/log.py", line 36, in
> >wrapper
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     return
> >method(*args, **kwargs)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/neutron/common/rpc.py", line 175, in
> >call
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     context, msg,
> >rpc_method='call', **kwargs)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/neutron/common/rpc.py", line 201, in
> >__call_rpc_method
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     return
> >func(context, msg['method'], **msg['args'])
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/rpc/client.py", line
> >389, in call
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     return
> >self.prepare().call(ctxt, method, **kwargs)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/rpc/client.py", line
> >152, in call
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     retry=self.retry)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/transport.py", line 90,
> >in _send
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     timeout=timeout,
> >retry=retry)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqpdriver.py",
> >line 408, in send
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     retry=retry)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqpdriver.py",
> >line 397, in _send
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     result =
> >self._waiter.wait(msg_id, timeout)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqpdriver.py",
> >line 285, in wait
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     reply, ending =
> >self._poll_connection(msg_id, timeout)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager   File
> >"/usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqpdriver.py",
> >line 235, in _poll_connection
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager     % msg_id)
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager MessagingTimeout: Timed
> >out waiting for a reply to message ID e928ae87e89e442790b84f053a75f58f
> >2015-01-22 16:10:52.712 14795 TRACE
> >neutron.services.loadbalancer.agent.agent_manager
> >
> >Digging into this, I see that it is running the v1 namespace_driver (not
> >synchronous_namespace_driver), so I edited the lbaas-agent.conf to load
> >synchronous_namespace_driver), and when I do that it fails to load
> >because __init__() takes two arguments instead of three.
> >
> >I'm clearly missing something here, does anyone have any suggestions?  I
> >would appreciate any advice, and can provide more details as needed.
> >
> >Thanks,
> >
> >Al
> >
> >
> >
> >
> >
> >
> >
> >
> >__________________________________________________________________________
> >OpenStack Development Mailing List (not for usage questions)
> >Unsubscribe:
> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> >http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>



-- 
Kevin Benton
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150123/36493469/attachment.html>