Operating System - HP-UX
1825160 Members
2346 Online
109679 Solutions
New Discussion

DNS latency problem with Bind.

 
Mark2010_1
Frequent Advisor

DNS latency problem with Bind.

Hi,
I have an HP server running Bind as master and authoritative DNS (Bind 9.2).
I have set two machines Linux and windows XP as dns server doing only forwarding and cache (proxy) upon Bind 9.4.2, on both machines I have in named.conf

zone "lab.net" in {
type forward;
forward only;
forwarders { 10.10.10.2; 10.10.10.3;};
};

Both machines have on /etc/resolv.conf

127.0.0.1
10.10.10.2
10.10.10.3

==> The problem I have is:
The first dig takes too much time 2 seconds. All the machines are on the same lab and the ping is ~ 2 ms.
The first dig can be done on two ways:

dig domain @127.0.0.1 => Very slow 2 s !
dig domain @10.10.10.2 => 1 ms !

After that, the cache is working fine ~ 1 ms. Other dig on the same domain is OK.

Bellow a named trace on the XP machine:

Thanks for any hint

IN A +
24-avr.-2008 14:48:36.234 client 127.0.0.1#1240: query (cache) host10.lab.net
/A/IN' approved
24-avr.-2008 14:48:36.234 client 127.0.0.1#1240: replace
24-avr.-2008 14:48:36.234 clientmgr @00BC3048: createclients
24-avr.-2008 14:48:36.234 clientmgr @00BC3048: recycle
24-avr.-2008 14:48:36.234 createfetch: host10.lab.net A
24-avr.-2008 14:48:36.234 client @00C36468: udprecv
24-avr.-2008 14:48:36.234 client 127.0.0.1#1240: next
24-avr.-2008 14:48:36.234 client 127.0.0.1#1240: request failed: duplicate query

24-avr.-2008 14:48:36.234 client 127.0.0.1#1240: endrequest
24-avr.-2008 14:48:36.250 fctx 01C89340(host10.lab.net/A'): timeout
24-avr.-2008 14:48:36.250 fctx 01C89340(host10.lab.net/A'): try
24-avr.-2008 14:48:36.250 fctx 01CC9348(./NS'): timeout
24-avr.-2008 14:48:36.250 fctx 01C89340(host10.lab.net/A'): query
24-avr.-2008 14:48:36.250 fctx 01CC9348(./NS'): try
24-avr.-2008 14:48:36.250 resquery 01C90D00 (fctx 01C89340(cvvms83.vms.ipc.sfr.n
et/A)): send
24-avr.-2008 14:48:36.250 fctx 01CC9348(./NS'): query
24-avr.-2008 14:48:36.250 fctx 01C89340(host10.lab.net/A'): too many ti
meouts, disabling EDNS0
24-avr.-2008 14:48:36.250 resquery 01CCFD08 (fctx 01CC9348(./NS)): send
24-avr.-2008 14:48:36.250 resquery 01C90D00 (fctx 01C89340(host10.lab.net /A)): sent
24-avr.-2008 14:48:36.250 fctx 01CC9348(./NS'): too many timeouts, disabling EDN
S0
….
1 REPLY 1
Derek Whigham_1
Trusted Contributor

Re: DNS latency problem with Bind.

I think this could be a problem with the search order , I looks like it is looking for a local DNS server first.

In SAM , modify the local search to be file i.e /etc/hosts then remove the 127.0.0.1 from /etc/resolv.conf

Can remember exactly where it is but it is in SAM
Divide and Conquer