Ask Your Question

VMs do not get IP

asked 2014-05-16 04:32:14 -0500

davide gravatar image


First time posting here!

Here's my problem (I am sure you guys can help me): I have installed Havana on fc20 using packstack (1 controller/compute + 1 compute). Installation is OK (Neutron VLAN) and everything seems to work. However, when I create a VM (either on controller or compute) that may or may not get an IP address. For example:

[fedora@controller ~(keystone_admin)]$ nova list

| ID                                   | Name        | Status | Task State | Power State | Networks              |
| ff3a1648-6112-436d-97e2-ba241f806587 | compute     | ACTIVE | None       | Running     | int_net= |
| 4d4fb9bb-c9a5-4b56-9cc1-59a6455f01f9 | compute2    | ACTIVE | None       | Running     | int_net= |
| 28090fbc-665c-4645-8957-de2a11920acd | controller  | ACTIVE | None       | Running     | int_net= |
| 07bf8e8b-6ecc-4af8-8d59-c6bb91230f16 | controller2 | ACTIVE | None       | Running     | int_net= |

VMs are up and running and the ports are created (and ideally those IPs assigned). However,/var/lib/neutron/dhcp/45ded610-f874-4ceb-81ae-87684db1178f/host shows

[fedora@controller ~(keystone_admin)]$ cat /var/lib/neutron/dhcp/45ded610-f874-4ceb-81ae-87684db1178f/host 

(sometimes also shows duplicates). In the case above controller 1 gets the correct IP. Compute 1&2 should get it (but they don't) and controller2 does not get IP. I saw cases where VMs running on the compute node got the correct IP.

I have checked all the bugs relative to IP assignment and the correspondent patches seem to be already in the code I am using. Any idea?

Many thanks!

edit retag flag offensive close merge delete



Could you point what repos for Havana you have been using for this setup on F20 ?

dbaxps gravatar imagedbaxps ( 2014-05-16 04:37:18 -0500 )edit

$ neutron net-list
$ ip netns | grep your_private-network-id
$ip netns exec qdhcp- your_private-network-id ifconfig
It will report you interface name like ns-xxxxxxxxx
Start one of your Vms and run
$ ip netns exec qdhcp- your_private-network-id tcdump -ln -i ns-xxxxxxxxx
for 5-7 min

dbaxps gravatar imagedbaxps ( 2014-05-16 04:49:31 -0500 )edit

Done. I spawned 2 VMs (1 on controller, 1 on compute) none of them got an IP.

Tcpdump shows

[fedora@controller ~(keystone_admin)]$ sudo ip netns exec qdhcp-45ded610-f874-4ceb-81ae-87684db1178f tcpdump -ln -i tap2721c71e-59
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on tap2721c71e-59, link-type EN10MB (Ethernet), capture size 65535 bytes
10:55:21.501558 IP6 :: > ff02::16: HBH ICMP6, multicast listener report v2, 1 group record(s), length 28
10:55:21.910347 IP6 :: > ff02::16: HBH ICMP6, multicast listener report v2, 1 group record(s), length 28
10:55:22.134052 IP > BOOTP/DHCP, Request from fa:16:3e:6e:7d:60, length 300
10:55:22.221090 IP6 :: > ff02::1:ff6e:7d60: ICMP6, neighbor solicitation, who has fe80::f816:3eff:fe6e:7d60, length 24
10:55:23.221839 IP6 fe80::f816:3eff:fe6e:7d60 > ff02 ...
davide gravatar imagedavide ( 2014-05-16 05:04:54 -0500 )edit

Hi dbaxps, thanks for your quick reply!

I did not use any particular repo, I just used the default fedora repos. Should I use a particular one?


davide gravatar imagedavide ( 2014-05-16 05:07:50 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2014-05-16 05:16:59 -0500

dbaxps gravatar image

Please , view .
I am afraid you have to follow instructions on this page to get systems up and running

edit flag offensive delete link more


Thanks dbaxps, I will give it a try.

But I also think it would be useful for the community to understand what's going on...

davide gravatar imagedavide ( 2014-05-16 05:31:49 -0500 )edit

Could you be so kind and upload the final version of your answer file to some location on the Net. I mean the one you ran second for deployment.

dbaxps gravatar imagedbaxps ( 2014-05-16 05:42:18 -0500 )edit

What's going on is not a secret, it's publicly available link 2014-May Archive by Author

dbaxps gravatar imagedbaxps ( 2014-05-16 05:47:03 -0500 )edit

HI, I have successfully installed Icehouse following your link (thanks!). The answer file I have used is ans.txt. Please note that is the controller and is the compute node. All services seem to work, however I can't boot an instance. I guess it is related to this bug

davide gravatar imagedavide ( 2014-05-19 03:51:47 -0500 )edit

@davide, please view (answer file is inside)
Two Node (Controller+Compute) IceHouse Neutron OVS&VLAN Cluster on CentOS 6.5
Two KVMs ,each one with 2 virtual NICs, have been used for test. I do can launch Ubuntu 14.04 and CirrOS instances and logged into via ssh, however network inside VMs is pretty slow. I want to repeat test on real boxes.SELINUX set to permissive here and bellow
Answers files look similar, but I didn't set

dbaxps gravatar imagedbaxps ( 2014-05-19 04:04:18 -0500 )edit

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



Asked: 2014-05-16 04:32:14 -0500

Seen: 164 times

Last updated: May 16 '14