Ask Your Question
1

Instances don't get an IP address assigned

asked 2014-10-21 10:22:23 -0500

ooms gravatar image

updated 2014-10-22 05:20:02 -0500

OpenStack Icehouse 3-node setup on CentOS 6.5 using Neutron, ML2 and OVS following the official install docs ( http://docs.openstack.org/icehouse/in... ).

controller, network and compute1 are virtual machines running under ESXi 5.5 attached to a single vSwitch which had been tagged for VLAN 4095. The vSwitch had been configured in promiscous mode, allowed forged transits and MAC address changes.

Hi. I'm stuck on the final piece of putting this Icehouse environment together. I've set up the internal and external networks, I believe, correctly, but the launched instances (tried cirros and centos) do not receive an IP address.

  • If I log into the instance from the console and set the IP address it should have been assigned by the DHCP agent, the instance can then ping the internal gateway.
    ifconfig eth0 125.125.0.1
  • If I add a default route within the instance, it can then ping the external gateway route add -net 0.0.0.0 125.125.0.1 eth0
  • If I assign a floating IP to the instance I can ssh on to the instance from the external network

Throughout the launching of the instances, the DHCP agent's log is not updated at all (verbose = yes) so I don't have much to go on. Everything seems to be fine... nova service-list, neutron agent-list, cinder, glance, etc all working.

/var/log/neutron/dnsmasq.log is empty and the output from addn_hosts is on pastebin. And some more details on my networks.

Any ideas where I should look to troubleshoot this please?

server.log output is here http://pastebin.com/Q1UB5gaJ

openvswitch-agent.log

2014-10-22 10:44:07.652 2238 INFO neutron.common.config [-] Logging enabled!
2014-10-22 10:44:09.386 2238 INFO neutron.openstack.common.rpc.impl_qpid [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Connected to AMQP server on controller:5672
2014-10-22 10:44:09.473 2238 INFO neutron.openstack.common.rpc.impl_qpid [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Connected to AMQP server on controller:5672
2014-10-22 10:44:12.883 2238 INFO neutron.plugins.openvswitch.agent.ovs_neutron_agent [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Agent initialized successfully, now running... 
2014-10-22 10:44:12.890 2238 INFO neutron.plugins.openvswitch.agent.ovs_neutron_agent [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Agent out of sync with plugin!
2014-10-22 10:44:13.004 2238 INFO neutron.plugins.openvswitch.agent.ovs_neutron_agent [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Agent tunnel out of sync with plugin!
2014-10-22 10:44:13.010 2238 INFO neutron.openstack.common.rpc.impl_qpid [req-b64b9fa7-5e53-43f2-89bb-b36ac47df782 None] Connected to AMQP server on controller:5672
2014-10-22 10:44:13.648 2238 INFO neutron.agent.securitygroups_rpc [-] Preparing filters for devices set([u'59ace705-bf5d-4b84-9a74-5ca293be691b', u'8edf09fb-64fb-41ec-b0f0-9cff0cc88cc6'])
2014-10-22 10:44:14.114 2238 WARNING neutron.plugins.openvswitch.agent.ovs_neutron_agent [-] Device 59ace705-bf5d-4b84-9a74-5ca293be691b not defined on plugin
2014-10-22 10:44:14.505 2238 INFO neutron.plugins.openvswitch.agent.ovs_neutron_agent [-] Port 8edf09fb-64fb-41ec-b0f0-9cff0cc88cc6 updated. Details: {u'admin_state_up': True, u'network_id': u'135185ea-a416-463b-a331-e9930551d42e', u'segmentation_id': 2, u'physical_network': None, u'device': u'8edf09fb-64fb-41ec-b0f0-9cff0cc88cc6', u'port_id': u'8edf09fb-64fb-41ec-b0f0-9cff0cc88cc6', u'network_type': u'gre'}
2014-10-22 10:44:14.505 2238 INFO neutron.plugins.openvswitch.agent.ovs_neutron_agent [-] Assigning 1 as ...
(more)
edit retag flag offensive close merge delete

Comments

1

Check dnsmasq running for each qdhcp-namespace && /var/log/neutron/dnsmasq.log
Also file /var/lib/neutron/dhcp/xxxxxxxxx/addn_hosts from dnsmasq command line string

dbaxps gravatar imagedbaxps ( 2014-10-21 10:26:09 -0500 )edit
1

The number of dnsmasq procces is the number of tenants vlans allocated . http://pastebin.com/qe9b8ssV point you to corresponding private vlan ( IP address interval).

dbaxps gravatar imagedbaxps ( 2014-10-21 10:52:30 -0500 )edit
1

Try decrease dhcp_lease_time=120 in nova.conf and run tcpdump -vv -i tap-interface of corresponding ip netns qdhcp-namespace ifconfig. It should capture dhcp requests from VM and dhcp responses from dnsmasq.

dbaxps gravatar imagedbaxps ( 2014-10-21 10:54:28 -0500 )edit

The tap device that the namespace is referring to does not seem to exist so I can't run tcpdump against it, as advised. Not sure where it's gone wrong during the install process. http://pastebin.com/kg99ywDJ

ooms gravatar imageooms ( 2014-10-21 15:31:02 -0500 )edit
1

If you edit your question as you debug the issue the question gets clearer.

smaffulli gravatar imagesmaffulli ( 2014-10-21 19:13:59 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
1

answered 2014-10-22 05:47:54 -0500

dbaxps gravatar image
edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

2 followers

Stats

Asked: 2014-10-21 10:22:23 -0500

Seen: 6,731 times

Last updated: Oct 22 '14