BIND 8.2.2-P5 is a pain!

Richard Smith rdls at
Sun Jan 23 18:05:40 GMT 2000

jon wrote:
> =

> On 22-Jan-00 Richard Smith wrote:
> > I've just upgraded to 3.4-RELEASE which upgraded BIND to the infamous=

> > 8.2.2-P5.
> >
> > In the past, I've managed to get bind, ppp, and sendmail to live happ=
> > together and only use the phone when I told them to. That is:- ppp wo=
> > dial on demand, but bind would satisfy all sendmails dns requests dur=
> > boot time using local name tables.
> >
> > Not with 8.2.2-P5, this version of bind insists on pinging the uplink=

> > nameservers on startup. I've looked at the ISC mailing list and the
> > general response is "tough, that's how it's done now".

I've fixed the problem by using the 8.1.2 binary off the 3.3 CD #2.


> Funny, I've just gotten the same problem - I put the
> 'ppp -alias -auto demon' line in my rc.conf (is this the best place?)
> but when I boot the machine, it dials then cuts the connection almost
> straight away.

I think you may be confused as to the purpose of rc.conf. It is a list
of environment variables, which are included several times within the
various rc.* files during boot time. It shouldn't invoke programs
directly. Loads of default values are stored in /etc/defaults/rc.conf,
which are then overridden by the values you set in /etc/rc.conf. You
should never need to change /etc/defaults/rc.conf as this is updated
every time the operating system is upgraded.

That said, I start ppp from my rc.conf file using the following


This will cause the following command to be executed at the appropriate
time during the boot up sequence and will hopefully avoid the false
starts your experiencing.

ppp -auto -nat -quiet papchap

Note that -alias is and alias for -nat and you can change papchap to
demon by using the following statement in your rc.conf:



> Another annoying thing is connections that just don't want to die.
> After I visit eg. Hotmail my server and one of their servers insist on
> bouncing packets back and forth so the line can take quite a while
> to drop.

I have seen some talk about this in the past, and apparently there is a
correct way to close a connection, but with some connections this
doesn't happen, perhaps because the dial-up line gets dropped. But TCP
has some long timers, being _very_ resilient to delay, and after a
minute or so will drop the connection unilaterally.


> Since I intend the machine to be autodialing when I'm not here I
> want to be sure it won't run up a huge 'phone bill for me.  Is there
> any way to stop this frivolous packet-bouncing?

Firstly, you should keep records so that you can limit your exposure to
huge phone bills. If you allow it to dial up during peak time, and
someone sends you a huge e-mail (most users don't even know they're
doing it sometimes), and the punts are running slow that day, you could
clock up =A320 or so bill during the day with your end working just fine.=

You need to decide what rules you want to impose then see how you can
achieve them. =

Also you might want to look at the various ppp filters, which determine
which types of packets can cause a dial-up and which types of packets
can hold the line up. I haven't used these (yet) so I can't advise.


More information about the Ukfreebsd mailing list