Ask Your Question

Packstack complains of not being able to detect ipaddress of interface

asked 2015-08-10 18:32:53 -0500

bnordgren gravatar image

updated 2015-08-14 09:40:58 -0500

rbowen gravatar image

About a month after installing RDO Kilo on CentOS 7 with packstack, I wanted to set up horizon to use https. So using the same answer file that we used originally, I edited the CONFIG_HORIZON_SSL option to read "y" instead of "n" and re-ran packstack, with the following result:

Adding Neutron FWaaS Agent manifest entries          [ DONE ]
Adding Neutron LBaaS Agent manifest entries          [ DONE ]
Adding Neutron API manifest entries               [ ERROR ]

ERROR : "Couldn't detect ipaddress of interface ten on node"
Please check log file /var/tmp/packstack/20150810-170722-80OEvA/openstack-setup.log for more information

The log file adds nothing to this error report. It is looking for interface "ten" (short for tenant) because:

# Interface for the Open vSwitch tunnel. Packstack overrides the IP
# address used for tunnels on this hypervisor to the IP found on the
# specified interface (for example, eth1).

However, you can:

[root@plume ~(keystone_admin)]# ssh
Last login: Mon Aug 10 16:54:15 2015 from
[root@compute-0-3 ~]# ip addr show dev ten
3: ten: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
   link/ether 00:1e:67:b3:75:42 brd ff:ff:ff:ff:ff:ff
    inet brd scope global ten
       valid_lft forever preferred_lft forever
     inet6 fe80::21e:67ff:feb3:7542/64 scope link
       valid_lft forever preferred_lft forever
[root@compute-0-3 ~]# ping
PING ( 56(84) bytes of data.
64 bytes from icmp_seq=1 ttl=64 time=0.267 ms
64 bytes from icmp_seq=2 ttl=64 time=0.196 ms
--- ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 999ms
rtt min/avg/max/mdev = 0.196/0.231/0.267/0.038 ms
[root@compute-0-3 ~]# logout
Connection to closed.
[root@plume ~(keystone_admin)]# ping
PING ( 56(84) bytes of data.
64 bytes from icmp_seq=1 ttl=64 time=0.231 ms
64 bytes from icmp_seq=2 ttl=64 time=0.187 ms
--- ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1000ms
rtt min/avg/max/mdev = 0.187/0.209/0.231/0.022 ms

I also attempted to run packstack using the original (unmodified) answer file. Same result. Could someone point me to why this answer file might be behaving this way this month, when last month it seemed happy?

edit retag flag offensive close merge delete


I have the same issue when I try to add a second compute node:

Adding Neutron API manifest entries [ ERROR ]

ERROR : "Couldn't detect ipaddress of interface eth1 on node X

Viktor gravatar imageViktor ( 2015-08-18 04:44:19 -0500 )edit

My answer file lacks a "HOST DETAILS" section. I don't know the format of this section, but it seems to be the only thing which can cause that error.


bnordgren gravatar imagebnordgren ( 2015-08-21 09:06:51 -0500 )edit

It is a bug, please report it.

mpa360 gravatar imagempa360 ( 2015-11-28 08:12:14 -0500 )edit

3 answers

Sort by ยป oldest newest most voted

answered 2015-10-04 07:45:16 -0500

sanderv gravatar image

I had the same problem. It was because packsack has integrated the eth1 device in the br-eth1 device, which now holds the IP address. There is a very simple quick-fix for this. Just use ip addr add dev eth1 to add a temporary IP address. Nobody cares which IP address this is, it will disappear after the packstack procedure, but allow you to add the node.

edit flag offensive delete link more


I like this better than downgrading.

bnordgren gravatar imagebnordgren ( 2015-11-09 19:14:38 -0500 )edit

answered 2015-08-20 18:15:30 -0500

bnordgren gravatar image

Further investigation revealed that packstack had been upgraded in the interim (via yum upgrade on the controller). I had to downgrade packstack so that the version which originally generated the answer file is the same as the version I'm using now.

[root@plume recovery]# repoquery --show-duplicates openstack-packstack*

In my case, this is the 0.1 version listed above. I downgraded all three packages to 0.1. Be a little careful, the 0.8 and the 1.0 versions were released the same day so downgrading to 0.8 did nothing for me. After downgrading to 0.1, I stopped getting the messages and packstack ran to completion.

Another possible solution would be to generate a new answer file with the new version of packstack, and edit it so that the configuration was the same as the original.

edit flag offensive delete link more

answered 2017-03-12 08:43:16 -0500

I have seen this issue in 3 different scenarios:Solution to each of them is as follows:

  1. if eth1 is the interface, then go to /etc/sysconfig/network-scripts and edit vi ifcfg-eth1. Set onboot to yes if it is no. and then ifup eth1

  2. The second time I saw this was when I was adding a compute to my existing packstack configuration. I had the controller added to EXCLUDE_SERVER list. Remove this from the list and it was working.

  3. As eth1 is added as a ovs port to br-eth1 bridge, need to statically add ip to the eth1 interface.

I saw these issues in Ocatta release.

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

1 follower


Asked: 2015-08-10 18:32:53 -0500

Seen: 1,515 times

Last updated: Mar 12