Description of problem: if neutron agent on host dies there is no log nor GUI event to notify the user about the fact Version-Release number of selected component (if applicable): Red Hat Enterprise Virtualization Manager Version: 3.4.0-0.21.el6ev additonal info: neutron agent was not launched properly on host during its installation to rhevm, I had no clue that something was wrong, everything was looking OK, this should not happen to our users [root@dell-r210ii-06 ~]# service neutron-linuxbridge-agent status neutron-linuxbridge-agent dead but pid file exists root@dell-r210ii-06 ~]# cat /var/log/neutron/linuxbridge-agent.log 2014-05-23 09:31:50.824 32139 INFO neutron.common.config [-] Logging enabled! 2014-05-23 09:31:50.824 32139 INFO neutron.plugins.linuxbridge.agent.linuxbridge_neutron_agent [-] Interface mappings: {} 2014-05-23 09:31:50.954 32139 INFO neutron.plugins.linuxbridge.agent.linuxbridge_neutron_agent [-] RPC agent_id: lb90e2ba0428c0 2014-05-23 09:31:50.959 32139 CRITICAL neutron [-] No module named openstack.common.rpc.impl_qpid 2014-05-23 09:31:50.959 32139 TRACE neutron Traceback (most recent call last): 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/bin/neutron-linuxbridge-agent", line 10, in <module> 2014-05-23 09:31:50.959 32139 TRACE neutron sys.exit(main()) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/plugins/linuxbridge/agent/linuxbridge_neutron_agent.py", line 981, in main 2014-05-23 09:31:50.959 32139 TRACE neutron root_helper) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/plugins/linuxbridge/agent/linuxbridge_neutron_agent.py", line 780, in __init__ 2014-05-23 09:31:50.959 32139 TRACE neutron self.setup_rpc(interface_mappings.values()) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/plugins/linuxbridge/agent/linuxbridge_neutron_agent.py", line 825, in setup_rpc 2014-05-23 09:31:50.959 32139 TRACE neutron consumers) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/agent/rpc.py", line 43, in create_consumers 2014-05-23 09:31:50.959 32139 TRACE neutron connection = rpc.create_connection(new=True) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/openstack/common/rpc/__init__.py", line 96, in create_connection 2014-05-23 09:31:50.959 32139 TRACE neutron return _get_impl().create_connection(CONF, new=new) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/openstack/common/rpc/__init__.py", line 306, in _get_impl 2014-05-23 09:31:50.959 32139 TRACE neutron _RPCIMPL = importutils.import_module(impl) 2014-05-23 09:31:50.959 32139 TRACE neutron File "/usr/lib/python2.6/site-packages/neutron/openstack/common/importutils.py", line 59, in import_module 2014-05-23 09:31:50.959 32139 TRACE neutron __import__(import_str) 2014-05-23 09:31:50.959 32139 TRACE neutron ImportError: No module named openstack.common.rpc.impl_qpid 2014-05-23 09:31:50.959 32139 TRACE neutron
I am unsure I understand, moti, I am sure you will. I also do not like we are working in parallel on upstream bugs on downstream. downstream bugs should be specific to downstream or customers.
As we would like to focus on Open vSwitch and remove the Linux bridge plugin, can you please retest this with Open vSwitch and change this RFE for the openvswitch-agent? Thanks, Nir
neutron-openvswitch-agent is the same case [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status neutron-openvswitch-agent (pid 14496) is running... [root@dell-r210ii-05 ~]# kill -9 14496 [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status neutron-openvswitch-agent dead but pid file exists [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status neutron-openvswitch-agent dead but pid file exists
(In reply to Martin Pavlik from comment #3) > neutron-openvswitch-agent is the same case > > [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status > neutron-openvswitch-agent (pid 14496) is running... > [root@dell-r210ii-05 ~]# kill -9 14496 > [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status > neutron-openvswitch-agent dead but pid file exists > [root@dell-r210ii-05 ~]# service neutron-openvswitch-agent status > neutron-openvswitch-agent dead but pid file exists Thanks Martin
I believe that we should probably have a vdsm caps (or better, stats) hook that reports about the agents that are running in the Host. Then, the engine code that handles vdsStats can generate the appropriate event (probably refuse to schedule a VM with the external provider network to that faulty Host too).
Dan, who's going to own this on network team?
It makes sense to keep this on ydary until it receives a planning ack
and in my opinion, tightening the neutron integration should not be high on our agenda right now.
Planning has indicated this request is declined. You may appeal this decision by reopening this request.