SpamAssassin dns timeouts... why?!

[Richard D Alloway]

Hi!  I have been having loads of problems with spamassassin timing out during 
DNS lookups...

If I use

/usr/bin/spamassassin -D < /tmp/spamemail.txt

I see the correct IP used for the nameserver:

[16018] dbg: dns: name server: 192.168.1.1, family: 2, ipv6: 0

Then, I see that the lookups took a LONG time:

[16018] dbg: uridnsbl: select found 1 socks ready
[16018] dbg: uridnsbl: queries completed: 1 started: 2
[16018] dbg: uridnsbl: queries active: DNSBL=10 NS=3 at Tue Dec 5 15:52:34 2006
[16018] dbg: uridnsbl: waiting 2 seconds for URIDNSBL lookups to complete
[16018] dbg: uridnsbl: select found 1 socks ready
[16018] dbg: uridnsbl: queries completed: 1 started: 0
[16018] dbg: uridnsbl: queries active: A=2 DNSBL=10 NS=2 at Tue Dec 5 15:52:34 
2006
[16018] dbg: uridnsbl: select found 1 socks ready
[16018] dbg: uridnsbl: query for winbeam.com took 50 seconds to look up 
(multi.surbl.org.:winbeam.com)
[16018] dbg: uridnsbl: queries completed: 1 started: 0
[16018] dbg: uridnsbl: queries active: A=2 DNSBL=9 NS=2 at Tue Dec 5 15:52:35 
2006
[16018] dbg: uridnsbl: select found 1 socks ready
[16018] dbg: uridnsbl: query for winbeam.com took 50 seconds to look up 
(multi.uribl.com.:winbeam.com)
[16018] dbg: uridnsbl: queries completed: 1 started: 0
...etc...
[16018] dbg: uridnsbl: done waiting for URIDNSBL lookups to complete
[16018] dbg: uridnsbl: aborting remaining lookups

I've tried using dnsmasq and bind9 as caching-only nameservers dedicated to 
resolving and caching queries for my mail servers, and the cache is growing, 
but multiple "/usr/bin/spamassassin -D < /tmp/spamemail.txt" attempts on the 
same file still result in 30-70 sec lookup times in the uridnsbl portion of 
spamassassin.

Manual queries of the hosts ('winbeam.com.multi.uribl.com' for example) resolve 
(or fail to) instantly from either the cache box or the mail server.

Any idea what could be wrong?  I'm rapidly running out of ways to try to 
increase performance here.

Thanks!

-Rich