liberty: dhcp agent not updating dnsmasq hosts file without restart of agent

asked 2015-10-23 16:34:59 -0600

jkilborn gravatar image

updated 2015-10-23 16:35:30 -0600

Since I upgraded from kilo to liberty in our pre-production environment, I'm having a problem with the dhcp agent. When I create an instance, it assigns and IP from the dhcp pool, but doesn't add it to the dnsmasq hosts (addn_hosts) file. So when the instance boots and does it dhcp discover, dnsmasq doesnt see the mac address in the hosts file, so it logs "no address available", and the instance doesn't get an IP. If I bounce the neutron-dhcp-agent, it rebuilds the dnsmasq files, and my instance will get a dhcp address. Something is not updating file during normal instance creation, but I'm not sure where that code is. Has anyone seen this with liberty? Is it a bug? It was working find in kilo.


edit retag flag offensive close merge delete


that's weird, but i do see the same behavior on OS kilo +nova-network

All same, hosts file isn't updated until nova-network service is restarted

elenhil gravatar imageelenhil ( 2015-10-29 06:06:12 -0600 )edit

2 answers

Sort by ยป oldest newest most voted

answered 2015-10-28 08:57:10 -0600

dbaxps gravatar image

Reproduced on RDO Liberty.

1. Reboot box
2. Create new RouterNew ( for demo ) and new private subnet for demo
3. Attach interface to private on  RouterNew and gateway to public.
4. Bug should be gone at least on RDO
edit flag offensive delete link more


Thanks for the response. We are using a flat provider network, so we dont have an l3 router. Our network is created as neutron net-create MYNETWORK --shared --provider:network_type flat --provider:physical_network provider

Can you point me to this bug, so I can take a look?

jkilborn gravatar imagejkilborn ( 2015-10-28 16:22:30 -0600 )edit

I didn't mean official bug.

dbaxps gravatar imagedbaxps ( 2015-10-29 01:20:08 -0600 )edit

answered 2019-01-23 08:19:21 -0600

vgdub gravatar image

updated 2019-01-23 10:54:12 -0600

Hey guys @dbaxps @jkilborn I am facing the same issue in my environment , I have to the point same issue , whenever I restart the neutron-dhcp-agent.service it updates the dnsmasq hosts file from database and all new VMs get the IP.

The funny part with my environment is that this is only happening very randomly , so like I have 120 namespaces and among this very randomly any 1 or 2 tenant network reports this issue nad it's not always the same tenant it's very random. I am just not sure what the issue is neutron-dhcp-agent is not able to do that and just give away this. any pointers would be appreciated.

I am aware of the fact that once neutron creates the port , dhcp agent which is talking to database has to update it's namespace files/records , but I am just not sure why randomly someday this doesn't happen for 1 or 2 tenant network.

I have controller which hosts API services and Neutron server on separate node and multiple compute nodes. Please help.

"DHCPDISCOVER(tapXXX) fa:16:3e:ec:bd:ad "no address available" dnsmasq-dhcp[3633]: 3621066038 available DHCP subnet: 172.17.x.x/

Also every time when I bounce the neutron-dhcp-agent service it reloads the files in dhcp namespace and all works fine. It also happened with me that when I tried to bounce the neutron-dhcp-agent service it also rebooted the neutron server itself.

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2015-10-23 16:34:59 -0600

Seen: 865 times

Last updated: Jan 23 '19