Event ID source incorrect for BIND 8.2.3 NT port

Danny Mayer mayer at gis.net
Thu May 3 05:42:39 UTC 2001


At 01:20 AM 5/3/01, Joe Rhett wrote:
> > Yes, I understand all of that.  However, what do you tell them about when
> > they should do this? If I understand you correctly you provide them with a FAQ
> > which says, in effect, if A happens, then do B.  I'd like to know what A is in order
> > for the answer be to say: "type 'ndc restart'".
>
>Changes to named.conf. That is 100% of the problem I am trying to deal
>with at this moment, however I would prefer that we didn't have an option
>that did something different on NT than on Unix. So my request was phrased
>as a feature match.

         Well reload does that and any zone file changes.  restart both deals
with changes to named.conf, the zone files as well as clearing cache and
memory.

> >  Since I know that you've administered some major networks, maybe you
> > are willing to explain to me why you find it necessary to clear the cache on a
> > server from time to time. What problem do you encounter that requires you
> > to do this? I one thing that seems to be clear so far is that the restart clears
> > cache as a "side-effect" of restart. I run my NT systems for months at a time
> > without rebooting them, in spite of what people say about needing to do so
> > regularly. I don't recommend that people reboot their systems after installing BIND.
> > So why restart BIND?
>
>This is not related to my request, and I have no real opinion on clearing
>the cache. It's not an issue for me. I just wouldn't want an option on NT
>to do something different than the same option on Unix. If 'restart'
>happens to clear the cache (because the process was stopped and started) I
>wouldn't want it to do something different on NT just because neither I nor
>anyone else I can name wants to track OS-specific changes.
>
>You made a comment about implementing 'reconfig' -- that would satisfy my
>immediate goal as well.

         It turns out to be a tougher problem than I have time for right now.


>Note however than I can (in theory) name a time that someone would want to
>clear the cache. This isn't a problem I'm trying to solve at this time, and
>there may be better ways to solve it. I'm just giving you a 'what-if'
>because you asked for one.
>
>         There are a hell of a lot more inexperienced administrators operating
>         Internet-connected systems than used to be, or probably should be. Many
>         of them don't grasp the TTL issues, and make changes to web or mail
>         servers without lowering the TTL in advance. Once the problem is
>         diagnosed properly, the only way I know of in Bind 8 to resolve the
>         problem is to clear the cache. I used to have some good tricks, but 
>         these fell under 'security problems' and Paul 'fixed' them ;-(
>
>Yes, yes. Stupid people should suffer their mistakes. However it doesn't
>change that one needs to 'do' when a VP calls down because they can't log
>into their partner-venture's website, or get mail to them.

         That's why BIND 9 is much tougher on enforcing the RFC's. You
sometimes need to force people to fix their problems as they won't do it
otherwise. It's not just stupid people, it's people who are short on time and have
deadlines they have to meet. And it's getting worse and won't get better.

                 Danny



More information about the bind-workers mailing list