Forwarding queries from a private network

[Kevin Darcy]

I think this problem was bug #597: "we weren't priming the cache in
forward-only mode." It was fixed subsequent to BIND 8.1.2. Upgrade to
BIND 8.2.2-p5.


- Kevin

Federico Albesano wrote:

> Hi,
>
> sorry for asking similar question, but what is really the difference from
> SERVFAIL and NXDOMAIN in log messages like following?
>
> Oct  6 18:01:34 calmail named[2047]: XX /192.168.10.183/i.am/A
> Oct  6 18:01:34 calmail named[2047]: sysquery: findns error (SERVFAIL) on
> ns.come.to?
> Oct  6 18:01:34 calmail named[2047]: sysquery: findns error (SERVFAIL) on
> ns1.fortunecity.com?
>
> I have a BIND-8.1.2 (B1) on a Linux server on a private network 192.168.x.0
> with the following old-style named.boot (then converted with
> named-bootconf.pl):
>
> directory                               /var/named
> ;cache           .                      named.ca
> primary         0.0.127.in-addr.arpa    named.local
> secondary       10.168.192.in-addr.arpa 192.168.10.4    named.secondary.10
> secondary       11.168.192.in-addr.arpa 192.168.10.4    named.secondary.11
> forwarders      195.xxx.yyy.zzz 195.xxx.yyy.zzz 195.xxx.yyy.www
> 195.xxx.yyy.www
> options         forward-only
>
> This name server forwards, via a private router beetween 195.xxx.yyy.0 and
> 192.168.x.0, all queries to a BIND-4.9.3 (B2) server connected to Internet
> with official IP address.
> If I query B1 with:
>
> host -a i.am
>
> I get the answer but also I get the previuos messages logged.
> If I do the same query on B2, I get the answer and no error logged.
> With nslookup it seems all OK in ns.come.to and ns1.fortunecity.com!
> So I traced the query with debug level 3 on B1:
>
> datagram from [127.0.0.1].1400, fd 20, len 22
> XX /127.0.0.1/i.am/MX
> req: nlookup(i.am) id 42374 type=15 class=1
> req: found 'i.am' as 'i.am' (cname=0)
> wanted(0x400c39c8, IN MX) [IN NS]
> wanted(0x40106190, IN MX) [IN NS]
> wanted(0x400c39c8, IN MX) [IN NS]
> wanted(0x40106190, IN MX) [IN NS]
> wanted(0x40104dd0, IN MX) [IN A]
> finddata: added 0 class 1 type 15 RRs
> findns: 2 NS's added for 'i'
> ns_forw()
> nslookup(nsp=0xbffff510, qp=0x400d289c, "i.am")
> nslookup: NS "ns.come.to" c=1 t=2 (flags 0x2)
> sysquery(ns.come.to, 1, 1, 0, 0)
> findns: No root nameservers for class IN?
> sysquery: findns error (SERVFAIL) on ns.come.to?
> ns_freeqry(0x40113008)
> nslookup: NS "ns1.fortunecity.com" c=1 t=2 (flags 0x2)
> sysquery(ns1.fortunecity.com, 1, 1, 0, 0)
> findns: No root nameservers for class IN?
> sysquery: findns error (SERVFAIL) on ns1.fortunecity.com?
> ns_freeqry(0x40113008)
> nslookup: 0 ns addrs total
> evSetTimer(ctx 0x809d008, func 0x8056840, uap 0, due 970846906.000000000,
> inter 0.000000000)
> forw: forw -> [195.223.192.131].53 ds=4 nsid=3280 id=42374 -1ms retry 15sec
> free_nsp: ns.come.to rcnt 1
> free_nsp: ns1.fortunecity.com rcnt 1
> datagram from [195.223.192.131].53, fd 4, len 84
> qfindid(3280) -> 0x400d289c
> Response (USER NORMAL -) nsid=3280 id=42374
> rrextract: dname i.am type 15 class 1 ttl 300
> rrextract: dname i.am type 15 class 1 ttl 300
> rrextract: dname po.v3.com type 1 class 1 ttl 82543
> ignoring additional info 'po.v3.com' type A
> rrsetupdate: i.am
> rrsetcmp: no records in database
> db_update(i.am, 0x400c4194, 0x400c4194, 0, 031, 0x400ea008)
> db_update: adding 0x400c4194
> rrsetupdate: i.am 0
> db_update(i.am, 0x400c36f8, 0x400c36f8, 0, 031, 0x400ea008)
> db_update: adding 0x400c36f8
> rrsetupdate: i.am 0
> resp: got as much answer as there is
> send_msg -> [127.0.0.1].1400 (UDP 20) id=42374
> qremove(0x400d289c)
> unsched(0x400d289c, 42374)
> ns_freeqry(0x400d289c)
>
> ...but this seems like arab to me...
> Is it possible that B1 do not forward to B2 all queries? Why?
> Only some queries cause the errors logged.
>
> Thanks in advace...
> Federico Albesano