[335] in athena10
Re: Caching DNS for Athena 10
daemon@ATHENA.MIT.EDU (Quentin Smith)
Mon Jul 28 17:06:47 2008
Date: Mon, 28 Jul 2008 17:05:53 -0400 (EDT)
From: Quentin Smith <quentin@MIT.EDU>
To: Greg Hudson <ghudson@mit.edu>
cc: Kevin Chen <kchen@mit.edu>, athena10@mit.edu
In-Reply-To: <1217278695.12433.40.camel@error-messages.mit.edu>
Message-ID: <Pine.LNX.4.64L.0807281704450.10861@vinegar-pot.mit.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
I think there's no good reason to depend on MIT's nameservers. We might as
well just leave BIND in its default configuration, which is to resolve
names itself using the root servers. That will help with the latest
performance problems of Hesiod/DNS and will prevent all Athena 10 machines
from going down when MIT's DNS servers fail.
--Quentin
On Mon, 28 Jul 2008, Greg Hudson wrote:
> On Mon, 2008-07-28 at 16:48 -0400, Kevin Chen wrote:
>> I misread this in my reply.
>>
>> Do you mean the current Athena behavior of forwarding queries for mit.edu
>> to the DNS servers authoritative for mit.edu, or do you mean having a
>> caching nameserver that forwards all queries to the MIT DNS servers?
>
> The idea is to configure dnsmasq to forward queries to the MIT name
> servers instead of whatever's listed in resolv.conf. For BIND, the idea
> would be changing it from a recursive resolver to a forwarding cache
> using the MIT name servers.
>
> It's not clear there is a practical benefit; it's just a way of making
> the DNS configuration more MITish while we're in there. I will probably
> not choose to implement it if no one can think of a practical reason to
> do so.
>
>
>
