[Dnsmasq-discuss] dhcp entries not being removed from dnsmasq

Graeme Peterson gp at gridly.ca
Wed May 17 19:12:06 BST 2017

I found the following entry in /var/log/auth.log which shows neutron as
root (sudo) calling dhcp_release on IP:, MAC: 
fa:16:3e:8e:83:97 two minutes after the DHCPACK and two minutes before
the dnsmasq-dhcp log failure saying: "not using configured address because it is leased to fa:16:3e:8e:83:97". Looks to me like
Openstack is calling dhcp_release correctly. The dnsmask log is in a
folder named "34705fcf-4f9c-48eb-b0bc-ac5091e181c8", same as the arg to
ip netns exec without the "qdhcp-" prefix. So far it looks like a
dhcp_release/dnsmasq issue to my untrained eye. I have been able to work
around it somewhat by setting the max lease time to 30 seconds, although
that seems to be introducing other issues, more investigation required

./auth.log:May 16 16:55:21 my-ucs-69 sudo:  neutron : TTY=unknown ;
PWD=/var/lib/neutron ; USER=root ; COMMAND=/usr/bin/neutron-rootwrap
/etc/neutron/rootwrap.conf ip netns exec
qdhcp-34705fcf-4f9c-48eb-b0bc-ac5091e181c8 dhcp_release tapc5399cce-70 fa:16:3e:8e:83:97 

Thanks again,

On 2017-05-17 12:24, Graeme Peterson wrote:

> Hi all.
> Sorry if this issue has been discussed and resolved, I am new to the list. I tried to find it in the list, and came across this reference to the issue from Jan 2016:
> http://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2016q1/010273.html
> What I am seeing is with OpenStack Newton on Ubuntu 16.10 (4.8.0-49-generic), with force_dhcp_release=True in /etc/nova/nova.conf, using tcpdump on the netns for the relevant Openstack network, I see dnsmasq receives the dhcp request, issues an IP, and that IP should be released (Openstack should be calling dhcp_release, I need to figure out how to verify that it is or is not happening, however force_dhcp_release=True is explicitly set in /etc/nova/nova.conf) , but it seems like the dhcp entry isn't being entirely released. The odd thing is that when a new VM wants an IP, tcpdump shows the request coming in for an address, but no reply, and OpenStack thinks it got an IP - the same one that used to belong to the recently terminated VM - but there is no dhcp offer in the tcpdump, and the dnsmasq log shows:
> May 16 16:53:15 dnsmasq-dhcp[40394]: 3306068020 DHCPREQUEST(tapc5399cce-70) fa:16:3e:8e:83:97
> May 16 16:53:15 dnsmasq-dhcp[40394]: 3306068020 tags: tag0, known, tapc5399cce-70
> May 16 16:53:15 dnsmasq-dhcp[40394]: 3306068020 DHCPACK(tapc5399cce-70) fa:16:3e:8e:83:97 host-132-16-0-13
> ...
> ...
> ...
> May 16 16:57:12 dnsmasq-dhcp[40394]: 461988430 available DHCP subnet:
> May 16 16:57:12 dnsmasq-dhcp[40394]: not using configured address because it is leased to fa:16:3e:8e:83:97
> May 16 16:57:12 dnsmasq-dhcp[40394]: 461988430 DHCPDISCOVER(tapc5399cce-70) fa:16:3e:31:de:d3 no address available 
> I don't see a log entry for a release of, not sure if there should be one.
> Is this a known and hopefully fixed issue? Can I provide further info to help investigate it?
> Thanks,
> Graeme 
> _______________________________________________
> Dnsmasq-discuss mailing list
> Dnsmasq-discuss at lists.thekelleys.org.uk
> http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/attachments/20170517/6c5949d7/attachment.html>

More information about the Dnsmasq-discuss mailing list