Re: SRV records - when?

[Peter Koch <[e-mail]>]

Hi Mark,

> 	draft-andrews-http-srv-01.txt is available

nice draft. Wonder what the impact on larger name servers would look like,
especially during transition.

In section 2 you give this example (considered dnsop relevant):

      Non-default port specified:

         URI:      http://example.com:8080/
         SRV RR:   _http._tcp.example.com. SRV   10 1 80 host2.example.com.
         CNAME RR: example.com.            CNAME host1.example.com.
         A RRs:    host1.example.com.      A     10.0.0.1
                   host2.example.com.      AAAA  1080::8:800:200C:417A

      Connect to: 10.0.0.1 port 8080

While formally "example.com" need not be the name of a zone, in real life it
is and that's the expectation of the average DNS user, even emphasized by
the phrase "you are example.com" introducing the example in section 3.
So, this CNAME RR will be in clear conflict with SOA/NS RRs, a situation
you correctly declare "illegal" in an earlier paragraph. Unfortunately
cases like this happen "out there" every day and render the zone unusable.
So I'd rather see a warning here - or the CNAME RR altered, since it's not
necessary in this constellation. Instead it's worth another example showing
explicitly why SRV based solutions work where CNAME won't.

While at it: I'd probably mention that the SRV processing will have no
influence on what's sent in the http "Host:" header. 

Finally, what is the length of a "session"? If it is predetermined by the TTL
(actually which? the SRV's or the A/AAAA RRs'?), would this imply restrictions
on how to set the TTL values?

-Peter