bind 8.2.4: limiting used memory?

[Kevin Darcy]

As long as the discussion still includes a significant amount of
informational content about what BIND's
capabilities/limitations/strengths/weaknesses are, I see it as relevant to
this list/newsgroup, even if it is being presented in adversarial terms.
The recent discussion between Dan and I, for instance, touched on some of
the more obscure/advanced features of BIND like "forward first", stub
zones, chroot'ing issues, logging alternatives, etc. which quite plausibly
was of interest to a BIND afficionado.

If the discussion degenerates to predominantly personal attacks or "djbdns
sucks and here are the reasons why", with very little discussion of how
BIND does things better, then I would say, yes, that is off-topic for the
group.

As I've said before, this is bind-users, not bind-help. While most of the
threads here consist of folks trying to help each other with
BIND configuration/operation, I see it as having broader charter than
that. I consider any reasonably information-dense assessment of BIND's
features and/or lack of same, as relevant to the list/newsgroup.

Oh, and AFAIK "religious wars" have always existed in parallel with
"sharing information" on Usenet. Fundamentally, there really is no bright
line to be drawn between "information" and "opinion" anyway. (If you want
to dispute that point, please be aware that I have a Philosophy degree and
I know how to use it :-)


- Kevin


Michael Kjorling wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Aug 9 2001 00:00 -0000, D. J. Bernstein wrote:
>
> > Will Yadley writes:
> > > start flamewars
> >
> > Learn to read. I didn't start any of these discussions. Knowles did.
> > If his list actually exists, he should post it, so we can evaluate it
> > on its merits.
>
> I belive the thread was started by Michael Renzmann a few days ago, by
> asking how he should go about telling BIND how much memory it can use
> for its cache, or how much would be used, for use in a limited memory
> environment. See Message-ID
> <200108061607.f76G7Q632030 at mailgate3.cinetic.de>. Somehow it drifted
> over to BIND vs. djbdns - I didn't quite pay attention to where it
> went wrong.
>
> > > tends not to play nice with other software and is often confusing
> >
> > Try following the installation instructions. Upgrades from BIND are
> > explained in detail in http://cr.yp.to/djbdns/frombind.html.
> >
> > ---Dan
>
> I followed the installation instructions for qmail 1.03 (why not
> djbsmtp? Oh, you have your own protocol - what was its name again?)
> and I ended up with a system that was hardly at all usable. Before I,
> after many hours of configuration, managed to get the qmail
> configuration right (mostly by looking at the only other system I have
> access to which was running qmail, and that server is now shut down
> for other reasons) I lost most likely around 100-150 emails and got
> unsubscribed off several mailing lists because of qmail's "sorry, that
> domain isn't in my list of allowed rcpthosts" and other error
> messages.
>
> One can blame sendmail for being hard to configure (frankly, it's
> fairly simple using the m4's) but at least it works out of the box,
> and _accepts mail_ destined for the local computer with a _minimum_ of
> work. I only had to tell it that kjorling.com was a local domain and
> that was it. Plus it actually inserts some useful information in the
> headers. Nice if you want to track down spam, for example. I don't
> remember qmail doing that, but I remember being annoyed at the
> extremely limited Received: header information it provided (I didn't
> check the RFCs, but it's got to be at the edge of non-conformant.)
>
> The comp.protocols.dns.bind newsgroup (and the bind-users mailing list
> which it is gatewayed to - two-way) has become a place for religious
> wars over one software package (djbdns) against another (BIND). This
> does nothing to serve the Internet community nor does it help the
> people who actually have to wade through such crap in order to get a
> hold of what might be wrong with _their_ setup.
>
> I happen to like BIND, personally. And whether it's (roughly) 75% or
> 95% of the Internet name servers that are running it, it is still a
> considerable amount of servers. That amounts to a considerable amount
> of expertise available should something go wrong. If something really
> screws up with my DNS I have several persons in my surroundings who
> are very knowledgeable on the subject and are willing to help me. I
> have never seen a book "DNS and djbdns", however I do have a copy of
> "DNS and BIND", 4th edition, on my bookshelf. And of course I can get
> in touch with Nominum - the people who are actually _writing_ and
> _maintaining_ BIND, and purchase a support contract that fits my needs
> for the moment. Or some other company that provides BIND support.
>
> djbdns might have its merits - I am not going to argue that point. I
> am sure there are some occassions when djbdns, dnscache etc. are doing
> better than BIND. But I don't belive it is without reason reason that
> just about three quarters of the name servers (according to Dan J.
> Bernstein) to some 95% (was it Brad Knowles who cited that number?)
> are using BIND. All the root servers are running BIND, as far as I
> know.
>
> Can we stop fighting over one software package over another and
> instead concentrate on actually helping others? After all, Usenet was
> even _invented_ to share information - not to host religious wars.
>
> This newsgroup/mailing list is moderated. But so far the moderator
> just seems to be approving everything - even a blank "test" message
> got through to everyone here not long ago. I am a big fan of free
> speech, but perhaps it is time to pull the plug here and say that
> "this is allowed, and this is not"? In such a case I belive the rules
> should be worked out in cooperation between the most frequent posters
> here (none mentioned and none forgotten) and whoever is the moderator.
> Of course open for comments from anyone.