Query source port 53

[Nico Kadel-Garcia]

Barry Margolin wrote:
> In article <bmk0iq$16j2$1 at sf1.isc.org>,
> Nico Kadel-Garcia  <nkadel at comcast.net> wrote:
> 
>>Barry Margolin wrote:
>>
>>>It's never been used as the source port for zone transfers, not even by
>>>BIND 4.  That uses TCP, and has always used an ephemeral source port.
>>>Otherwise, you wouldn't be able to have multiple concurrent zone transfers
>>>between the same master and slave.
>>
>>
>>I'm not expert enough to know that you're completely incorrect, but am 
>>expert enough in networks to know that your reasoning is deeply flawed. 
> 
> 
> Sounds like you don't know much about the internal operation of TCP, though.

Various guts of it, a great deal. The formal language of "source port" 
vs. remote port that you described, no: I knew it once about 5 years ago 
and had lost track of it.

But the typical "port 53" is on the DNS server from which the zone 
transfers are pulled, right? It's the client pulling the zone that gets 
the response back on the ephemereal port, which is the remote port 
relative to that DNS server. The source port on the server itself is 
still 53. That's what I thought you were contradicting, and perhaps 
misunderstood what you meant.