Re: [Exim] Delivery optimization tips for large lists ?

Kezdőlap
Üzenet törlése
Válasz az üzenetre
Szerző: Pete Naylor
Dátum:  
Címzett: exim-users
Új témák: Re: [Exim] Delivery optimization tips for large lists ?
Tárgy: Re: [Exim] Delivery optimization tips for large lists ?
On Mon, 11 Oct 1999, Peter Radcliffe wrote:

> Pete Naylor <pete@???> probably said:
> > Nobody can help you if you misuse the software. You should configure it
> > appropriately for your needs.
>
> I expect things to be set up reasonably by default. The default time
> for caching negative hits was way too long.


For you perhaps. The default cannot be reasonable for everybody. The
default Solaris install is inappropriate for many needs too.

> I don't see any misuse in expecting reasonable defaults.


I see misuse if you knowingly fail to configure the software to suit your
particular environment.

> > That would be a nice feature to add, but it's not terribly important.
>
> *splutter*


Hmmm. That's convincing.

> > Please explain in more detail. I fail to see how round robin DNS is
> > related to providing multiple mail servers with any sort of failover
> > mechanism so I have to assume that this is another case of misuse.
>
> The MX record for example.com points at mail.example.com.
> mail.example.com has two A records, 10.0.0.1 and 10.0.0.2, two
> distinct machines for failover. nscd caches mail.example.com as
> 10.0.0.1 when someone sends mail to example.com. 10.0.0.1 goes down.
> Mail never reaches 10.0.0.2 until the first A record times out from
> nscd, which is a long time since nscd's default timeouts for positive
> caching are quite high.
>
> This is broken.


That's right - I hope nobody is silly enough to configure their DNS in the
manner that you've just described. Really bad practice.

> (Yes, I know that for this case a better solution is multiple MX
> records, this is just an example relavent to the list since some
> people do do it this way and nscd breaks it. Any system that uses
> failover with multiple A records is broken in this way with nscd)


Just as it is broken on machines where the resolver does not properly deal
with DNS round robin. Relying on this arrangement for redundancy is
nothing short of foolish. Can you provide an example which somebody in
their right mind might encounter?

> > They're very different applications though, and each has its pros and
> > cons. An ideal solution probably involves using both (though I wouldn't
> > necessarily run named on the same host).
>
> If you are only using DNS for name service they are the same
> application - making and caching DNS lookups. Just that nscd doesn't do
> it correctly.


nscd fulfills its purpose wonderfully when correctly configured. It is
also much closer to the application and more efficient than making every
request to a caching named (even one on the same machine).

> If it works for you, great.


It does, thanks. Works for many others too.

> It caused me horrible problems and until
> those problems are fixed I'm not going to use it,


Is everything else in the Solaris install preconfigured to suit you?

> but please don't be
> so uncouth as to insult me by claiming I'm abusing something so it
> doesn't work.


I never mentioned abuse, and if requesting that you back up your sweeping
statements about the software is uncouth, so be it. Personally, I think
it's a little rude to loudly proclaim someone else's work as crap without
considering all the variables or how it might better suit others.

-- 
/*-------------------------------------------------------------*
 * Peter James Naylor ## SysAdmin, Supernal Technologies, Inc. *
 *-------------------------------------------------------------*
       * pete@??? ## <http://www.supernal.net> *
       *------------------------------------------------*/