I am in the process of upgrading my web/mail/proxy/etc server from a
Deskpro PII/450 (named hermes) to a Netvista PIII/600 (named athena).
I am doing the transition gradually, by slowly moving the services over time.

Both machines run Debian testing+unstable. Both machines are connected
to the DSL switch (network 10.10.0.x) and to the internal gigabit
switch (network 10.0.0.x).

My problem is with the DNS server: right now both machines run bind
(bind9 9.3.1-2.0.1), with the same db.* except for the appropriate
records (each server thinks it is the authoritative server for both
10.0.0.x and 10.10.0.x). Both servers forward requests to the same
pair of DNS servers at the ISP (visi).

When I set the proxy server (squid 2.5.12-4) running on athena to use
the DNS server running on athena, I get 10-15 seconds delay in the web
browser. Watching the access.log, there is no entry for the URL I have
just requested. The URL will show after 10-15 seconds.

If I change the resolv.conf on athena to point to hermes, there is no
delay, and everyting is fast as expected. Only when athena is using
the local server, I can see the delay. The delay is there even if the
DNS server on hermes is down (no competition).

Both servers are unloaded, have 512 MB RAM, run Shorewall... They are
configured quite identical.

Does anybody have any idea what is going on or what should I do to try
to isolate the problem?

Thank you,
florin

-- 
Don't question authority: they don't know either!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://mailman.mn-linux.org/pipermail/tclug-list/attachments/20060213/19134753/attachment.pgp