Re: 'dig -t any ...' question

Jim Reid wrote:
>>>>>>"Sara" == Sara <demone33@yahoo.it> writes:
>
>
> Sara> Hi all, please take a look at the below reported commands I
> Sara> issued from DNS (BIND 9.2.1 on Red-Hat 7.3) and the output I
> Sara> got Briefly, I issued:
>
> Sara> # dig -t any ericsson.com
> Sara> # dig -t mx ericsson.com
> Sara> # dig -t any ericsson.com
>
> Sara> and I got different output for the dig -t any ericsson.com
> Sara> commands. So, which is the way the 'dig -t any' command
> Sara> works? Thank you all.
>
> There's nothing strange happening here. All you're seeing is routine
> cacheing and resolving behaviour. Your name sever knew nothing about
> ericsson.com. So the first lookup terminates when your name server
> retrieves the name servers for ericsson.com from the .com name
> servers.

jim, basically what you saying is that it answers with glue records, but
how come the glue is provided to the recursive client talking to the
recursive server, shouldn't it provide only *answers* or better than that??

how come the recursive server doesn't not even try to get better answer
for a 'recursion required' client, when the recursive server definitely
knows it is just a glue, or it doesn't ...? perhaps erricsson servers
don't not exist at all and nobody will ever find out after this single
dig since it doesn't even try to ask them...

Ladislav

> That info is cached by your server. On your next lookup,
> those ericsson.com servers get asked for MX records. These get cached
> by your server too and the reponse is relayed back to your lookup with
> dig. By the time of your last lookup, your name server has cached a
> bunch of stuff for ericsson.com, so your server just returns that.


>
> If you repeated that query a few minutes later, you'd see that the
> answers now had lower TTLs than before because they're on a countdown
> for expiry from your name server's cache. In fact you can see that in
> your posting. The MX related records have a 300 second TTL. But in the
> next lookup, the TTL was 298 seconds.
>