resolving problems

Hi, I've set up dnscache (on 192.168.0.1) and tinydns (127.0.0.1) for my
home toy network. /etc/resolv.conf says "nameserver 192.168.0.1". The
djbdns tools (dnsip, dnsipq, dnsname) all show that it's all up and
running properly, forward and reverse DNS lookups work.

BUT:

while ping is able to resolve a host name (served by tinydns), telnet
can't. While it takes a very long time for apache2 to start, squid
starts as normal (their Listen directive has the same host name). While
sshd takes ages to respond with the Password: prompt, the dictd
client/server works like a charm. What the hell is going on? All of
these problems go away if I simply put the relevant mappings into
/etc/hosts, but that's not what I want.

Please help, as I'm nearly desperate. Thank you very much in advance.
And yes, I'm running the latest Debian/unstable, glibc 2.3.2, kernel
2.6.0-test11.

"rihad" <(E-Mail Removed)> wrote in message
news:brmagf$1hc0q$(E-Mail Removed)

> while ping is able to resolve a host name (served by tinydns), telnet
> can't. While it takes a very long time for apache2 to start, squid
> starts as normal (their Listen directive has the same host name).
> While sshd takes ages to respond with the Password: prompt, the dictd
> client/server works like a charm. What the hell is going on? All of
> these problems go away if I simply put the relevant mappings into
> /etc/hosts, but that's not what I want.

Pare the name resolution down to the bare essentials, removing such spurious
entries as db, nisplus, nis. Have dns first in the order:

$ grep hosts /etc/nsswitch.conf
hosts: dns files


tony

--
use hotmail for any email replies



-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 100,000 Newsgroups - 19 Different Servers! =-----

ynotssor wrote:
> "rihad" <(E-Mail Removed)> wrote in message
> news:brmagf$1hc0q$(E-Mail Removed)
>
>
>>while ping is able to resolve a host name (served by tinydns), telnet
>>can't. While it takes a very long time for apache2 to start, squid
>>starts as normal (their Listen directive has the same host name).
>>While sshd takes ages to respond with the Password: prompt, the dictd
>>client/server works like a charm. What the hell is going on? All of
>>these problems go away if I simply put the relevant mappings into
>>/etc/hosts, but that's not what I want.
>
>
> Pare the name resolution down to the bare essentials, removing such spurious
> entries as db, nisplus, nis. Have dns first in the order:
>
> $ grep hosts /etc/nsswitch.conf
> hosts: dns files
>

Tried that to, to no avail. Besides, now that my /etc/hosts is basically
empty (nothing except an entry for localhost), chances are good the
order is irrelevant. And I'm not using nis, nisplus or whatever.

OK, here's updated info, I managed to trim the problem down to this:

# tcpdump -n host 192.168.0
tcpdump: listening on eth0

Now, when I do "ping rihad.localnet" I get instant name resolution:
12:16:57.789729 192.168.0.2.49154 > 192.168.0.1.53: 16599+ A?
rihad.localnet. (32)
12:16:57.790121 192.168.0.1.53 > 192.168.0.2.49154: 16599 1/0/0 A
192.168.0.1 (48) (DF)

But when I do "telnet rihad.localnet 80", name resolution hangs:
12:20:51.098895 192.168.0.2.49165 > 192.168.0.1.53: 16516+ AAAA?
rihad.localnet. (32)
12:20:56.106436 192.168.0.2.49166 > 192.168.0.1.53: 16516+ AAAA?
rihad.localnet. (32)
12:21:06.117134 192.168.0.2.49167 > 192.168.0.1.53: 16516+ AAAA?
rihad.localnet. (32)
12:21:26.128575 192.168.0.2.49168 > 192.168.0.1.53: 16516+ AAAA?
rihad.localnet. (32)
12:21:50.157479 192.168.0.1.53 > 192.168.0.2.49165: 16516 ServFail 0/0/0
(32) (DF)
12:21:50.157595 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
49165 unreachable (DF)
12:21:55.167227 192.168.0.1.53 > 192.168.0.2.49166: 16516 ServFail 0/0/0
(32) (DF)
12:21:55.167342 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
49166 unreachable (DF)
12:22:05.174194 192.168.0.1.53 > 192.168.0.2.49167: 16516 ServFail 0/0/0
(32) (DF)
12:22:05.174311 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
49167 unreachable (DF)
12:22:06.141472 192.168.0.2.49169 > 192.168.0.1.53: 16517+ A?
rihad.localnet. (32)
12:22:06.141791 192.168.0.1.53 > 192.168.0.2.49169: 16517 1/0/0 A
192.168.0.1 (48) (DF)

As you can see, only after more than a minute of trying does telnet's
resolver fall back to normal DNS A record query (the same thing ping
does right from the start) and eventually name resolution succeeds. Why
so? Aren't ping and telnet using the same resolver? What's the "AAAA?"
query anyway?

"rihad" <(E-Mail Removed)> wrote in message
news:brmgah$4tmv8$(E-Mail Removed)

> Now, when I do "ping rihad.localnet" I get instant name resolution:
> 12:16:57.789729 192.168.0.2.49154 > 192.168.0.1.53: 16599+ A?
> rihad.localnet. (32)
> 12:16:57.790121 192.168.0.1.53 > 192.168.0.2.49154: 16599 1/0/0 A
> 192.168.0.1 (48) (DF)
>
> But when I do "telnet rihad.localnet 80", name resolution hangs:
> 12:20:51.098895 192.168.0.2.49165 > 192.168.0.1.53: 16516+ AAAA?
> rihad.localnet. (32)
> 12:21:50.157479 192.168.0.1.53 > 192.168.0.2.49165: 16516 ServFail
> 0/0/0 (32) (DF)

> What's the "AAAA?"query anyway?

IPv6 queries, rather than IPv4. Your tinydns is misconfigured, I'm thinking.


--
use hotmail for any email replies



-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 100,000 Newsgroups - 19 Different Servers! =-----

rihad wrote:
> # tcpdump -n host 192.168.0
> tcpdump: listening on eth0
>
> Now, when I do "ping rihad.localnet" I get instant name resolution:
> 12:16:57.789729 192.168.0.2.49154 > 192.168.0.1.53: 16599+ A?
> rihad.localnet. (32)
> 12:16:57.790121 192.168.0.1.53 > 192.168.0.2.49154: 16599 1/0/0 A
> 192.168.0.1 (48) (DF)
>
> But when I do "telnet rihad.localnet 80", name resolution hangs:
> 12:20:51.098895 192.168.0.2.49165 > 192.168.0.1.53: 16516+ AAAA?
> rihad.localnet. (32)
> 12:20:56.106436 192.168.0.2.49166 > 192.168.0.1.53: 16516+ AAAA?
> rihad.localnet. (32)
> 12:21:06.117134 192.168.0.2.49167 > 192.168.0.1.53: 16516+ AAAA?
> rihad.localnet. (32)
> 12:21:26.128575 192.168.0.2.49168 > 192.168.0.1.53: 16516+ AAAA?
> rihad.localnet. (32)
> 12:21:50.157479 192.168.0.1.53 > 192.168.0.2.49165: 16516 ServFail 0/0/0
> (32) (DF)
> 12:21:50.157595 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
> 49165 unreachable (DF)
> 12:21:55.167227 192.168.0.1.53 > 192.168.0.2.49166: 16516 ServFail 0/0/0
> (32) (DF)
> 12:21:55.167342 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
> 49166 unreachable (DF)
> 12:22:05.174194 192.168.0.1.53 > 192.168.0.2.49167: 16516 ServFail 0/0/0
> (32) (DF)
> 12:22:05.174311 192.168.0.2 > 192.168.0.1: icmp: 192.168.0.2 udp port
> 49167 unreachable (DF)
> 12:22:06.141472 192.168.0.2.49169 > 192.168.0.1.53: 16517+ A?
> rihad.localnet. (32)
> 12:22:06.141791 192.168.0.1.53 > 192.168.0.2.49169: 16517 1/0/0 A
> 192.168.0.1 (48) (DF)
>
> As you can see, only after more than a minute of trying does telnet's
> resolver fall back to normal DNS A record query (the same thing ping
> does right from the start) and eventually name resolution succeeds. Why
> so? Aren't ping and telnet using the same resolver? What's the "AAAA?"
> query anyway?
>

OK, some googling helped. "AAAA" is an ipv6 address query. No wonder
telnet sucked. telnet -4 works just like ping.