query-source bug?

Joost Stegeman stegeman at xs4all.nl
Tue Feb 22 10:03:55 UTC 2000


Thanx Kevin,

I think it's the named-xfer that does a SOA query to determine the
current serial# with UDP and then an AXFR using TCP. In the source I
found the undocumented option '-x addr' which gives the option to
supply an IP address to bind to. I'll try that one later.

- Joost.

On 21 Feb 2000 17:25:04 -0800, Kevin Darcy <kcd at daimlerchrysler.com>
wrote:

>From the documentation of the "options" statement:
>
>
>> Note: query-source currently applies only to UDP queries; TCP queries always use a wildcard IP address and a random unprivileged
>> port.
>>
>
>- Kevin
>
>Joost Stegeman wrote:
>
>> Hi all,
>>
>> I'm a bit puzzled by the following behaviour of bind 8.2.5.
>> I have a setup where named is bound to a secondary (virtual) IP
>> address for both listening (listen-on) and querying (query-source).
>> After starting named, it listens to the correct address and starts
>> doing xfers from this address to port 53 of the remote host using UDP
>> 53.
>> As the remote host does not exist yet, it times out and stops trying.
>> This is expected behaviour AFAIK.
>>
>> BUT, after 5 tries, it starts doing TCP to port 53 of the remote host
>> from the boxes physical (=wrong) address, instead of the specified
>> query-source address. As the remote host doesn't exist, this is
>> signalled by a router sending "ICMP Time Exceeded (in transit)"
>> messages. After another 5 tries and a waiting period, the whole
>> process starts again.
>>
>> What's going on here, why are the TCP queries coming from the wrong
>> address? Is this a bug?
>>
>> BTW: box = Solaris 2.6
>>
>> tia, Joost.
>
>
>
>
>
>




More information about the bind-users mailing list