[Dnsmasq-discuss] IP6 DNS UDP strange behaviour
roger at beardandsandals.co.uk
Tue Jun 27 14:37:40 BST 2017
Sending ICMP unreachable is not very friendly with my ISP. They treat the
host as offline. The host is my VoIP PBX so I don't get any incoming calls
for a while. Would it not be better to hold the port open until all
requests have been received. I have turned on the strict-order option to
try and circumvent this.
On 27 June 2017 10:31:59 am Simon Kelley <simon at thekelleys.org.uk> wrote:
> What's probably happening is this.
> 1) query arrives from client,
> 2) UDP socket opened, and query send from socket to more than one
> upstream server.
> 3) Reply arrives from fastest upstream server, reply returned to client.
> 4) upstream socket closed.
> 5) reply arrives from slower upstream server -> destination unreachable.
> dnsmasq does the "send the query to all available servers" trick every
> once in a while to find the fastest one.
> If this is what you're seeing, it's a feature, not a bug.
> On 27/06/17 06:56, Roger James wrote:
>> I am seeing some rather perplexing behaviour regarding the reception of
>> upstream IPV6 UDP DNS query responses on a Debian (raspian) system
>> running on a Raspberry Pi. It appears that dnsmasq is not holding open
>> an IPV6 UDP port to handle a response to an upstream query. What my
>> network monitor shows is the upstream request going out and a reply
>> coming back in. However the Pi rejects the incoming UDP packet with a
>> ICMPV6 Destination unreachable (port unreachable).
>> I may be missing something obvious here, but it looks like dnsmasq is
>> not holding open a listening port for the response.
>> Has anyone got any ideas on what is going, or can point me at some ways
>> of tracking this down.
>> The configuration of dnsmasq is out of the box Debian.
>> Dnsmasq-discuss mailing list
>> Dnsmasq-discuss at lists.thekelleys.org.uk
> Dnsmasq-discuss mailing list
> Dnsmasq-discuss at lists.thekelleys.org.uk
More information about the Dnsmasq-discuss