Root and Forwarders Servers
Mark.Andrews at nominum.com
Mark.Andrews at nominum.com
Tue May 30 07:22:20 UTC 2000
The following bug is biting you. From the 8.2.3 change list.
910. [bug] address-sorting logic was exiting early.
In src/bin/named/ns_forw.c: nslookup change
if (n > 1) {
qsort((char *)qp->q_addr, n, sizeof(struct qserv),
(int (*)(const void *, const void *))qcomp);
}
to
/* Just sort the NS RR's we added, since the forwarders may
* be ahead of us (naddr > 0)
*/
if (n > naddr) {
qsort((char *)(qp->q_addr+naddr), n-naddr, sizeof(struct qserv),
(int (*)(const void *, const void *))qcomp);
}
This will work around the problem in qcomp.
Mark
>
> Hi there,
>
> I've just installed new version of BIND (Version = named 8.2.2-P5)
> instead of good old 4.9.7 at one of my servers and now doing some testing.
> So, I'd be thankfull if anybody could clarify some nuances of
> "forwarders" option:
>
> 0) I set "forwarders first" and "forwarders server" option and have
> default root.cache file.
>
> 1) When answering a query, named first tries all(?) root servers,
> and only after this tries my "forwarders server" 193.193.193.100.
> This significantly slows dns responces.
>
> The problem is that I expect named to query "forwarders server"
> first, not one of the root servers, because my forwarder answers much
> faster than root does. Nevertheless, I dont want to use
> "forwarders only " option, cause this will disable use of root servers
> even if my "forwarder" is unavailable.
>
>
> Some debug 1 output from named.run (with no connection to root servers):
>
> datagram from [client.ip].1266, fd 20, len 31
> req: nlookup(www.digex.net) id 1475 type=1 class=1
> req: found 'www.digex.net' as 'net' (cname=0)
> forw: forw -> [205.188.185.18].53 ds=4 nsid=61164 id=1475 16ms retry 4sec
> resend(addr=1 n=0) -> [195.8.99.11].53 ds=4 nsid=61164 id=1475 64ms
> resend(addr=2 n=0) -> [198.41.0.21].53 ds=4 nsid=61164 id=1475 93ms
> resend(addr=3 n=0) -> [198.41.3.38].53 ds=4 nsid=61164 id=1475 99ms
> resend(addr=4 n=0) -> [198.17.208.67].53 ds=4 nsid=61164 id=1475 118ms
> resend(addr=5 n=0) -> [192.5.5.241].53 ds=4 nsid=61164 id=1475 118ms
> resend(addr=6 n=0) -> [210.176.152.18].53 ds=4 nsid=61164 id=1475 252ms
> resend(addr=7 n=0) -> [198.41.0.4].53 ds=4 nsid=61164 id=1475 451ms
> resend(addr=8 n=0) -> [216.33.75.82].53 ds=4 nsid=61164 id=1475 971ms
> resend(addr=9 n=0) -> [192.112.36.4].53 ds=4 nsid=61164 id=1475 1006ms
> resend(addr=10 n=0) -> [192.36.144.253].53 ds=4 nsid=61164 id=1475 1419ms
> reforw(addr=11 n=0) -> [193.193.193.100].53 ds=4 nsid=61164 id=1475 -1ms
> ^^^^^^^^^^^^^^^^^
>
> resend(addr=12 n=0) -> [192.203.230.10].53 ds=4 nsid=61164 id=1475 2697ms
> resend(addr=0 n=1) -> [205.188.185.18].53 ds=4 nsid=61164 id=1475 16ms
> resend(addr=1 n=1) -> [195.8.99.11].53 ds=4 nsid=61164 id=1475 64ms
> datagram from [193.193.193.100].53, fd 4, len 123
> ^^^^^^^^^^^^^^^^
> send_msg -> [client.ip].1266 (UDP 20) id=1475
>
>
> BR,
>
> Bernt
>
>
>
>
>
--
Mark Andrews, Nominum Inc.
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: Mark.Andrews at nominum.com
More information about the bind-users
mailing list