[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Metadata service caching old nameservers?

On 8/29/19 6:26 AM, Grant Morley wrote:
> Hi All,
> We have a bit of a weird issue with resolv.conf for instances. We have 
> changed our subnets in neutron to use googles nameservers which is fine. 
> However it seems that when instances are launched they are still getting 
> the old nameserver settings as well as the new ones. If I look at the 
> metadata networking service it returns the old nameservers as well as 
> the new ones below:
> curl -i
> HTTP/1.1 200 OK
> Content-Type: text/plain; charset=UTF-8
> Content-Length: 753
> Date: Thu, 29 Aug 2019 09:40:03 GMT
> {"services": [{"type": "dns", "address": ""}, {"type": 
> "dns", "address": ""}, {"type": "dns", "address": 
> ""}, {"type": "dns", "address": ""}]
> In our neutron dhcp-agent.ini file we have the correct dnsmasq 
> nameservers set:
> dnsmasq_dns_servers =,
> Are there any database tables I can change or clear up to ensure the old 
> nameservers no longer get set? I can't seem to find any reference in our 
> config any more of the old nameservers, so I assume one of the services 
> is still setting them but I can't figure out what is.

So does the DHCP response on lease renewal just have the two Google 
nameservers in it?  If so, does a new VM booted on the subnet have the 
correct metadata values reported?  Just trying to narrow-down if this is 
in the neutron code or not.