G'day Kaz and thanks for your reply.
The diald doco specifically says to not use the -detach option and I am pretty
sure I had the same situation before I started using diald.
I will however, try your suggestion regarding signal level.
I have already tried killing diald (which in turn stops pppd) but this still
gave me interface ppp1 after restart.
Regarding my firewall rules, I allow my dialin users to do things such as
telnetting to other hosts on my net and initially set up the firewall rules
using -W ppp0 to restrict external traffic and -W ppp1/ppp2 for dialin users.
Given my restart problem, I have now rewritten the rules for dialin users to
use -S and -D but that still means that any rule with -W ppp0 does nothing
when pppd connects via ppp1.
Yes, my permanent connection is permanent (except for the aforementioned
failures) and I do own a /24 network.
Cheers and thanks,
Stephen.
> On Sat, 24 Oct 1998, Stephen Davies wrote:
>
> > Can anyone please explain why pppd might use ppp1 instead of ppp0 and better
>
> Because ppp devices are dynamically allocated as PPP line disciplines are
> installed on tty devices. When pppd starts up, it opens the /dev/tty*
> devic ethat it is told to use and then issues a special system call to install
> the PPP discipline on that tty. The ppp line discipline driver immediately
> kicks into action creates a network device called pppX, where X is the lowest
> number that is not currently in use. The pppX device will be released when the
> PPP line discipline on its corresponding tty device is removed; its name will
> then be available for reuse. This happens when the corresponding pppd daemon
> exits. Even if the daemon is forcibly terminated, the line discpline should be
> restored back to normal as the tty device is closed. (But if some other
> process is holding the tty open, the line discipline and hence the ppp device
> WON'T be released until the last process closes the tty!)
>
> Perhaps there is a race condition in your setup. I don't know how diald
> manages the pppd processes that it launches; but if the pppd sessions are
> detached, they are collected by init rather than whatever process launched
> them. [[ By deafult, pppd sets itself up as a deamon: it forks a child clone of
> itself and then terminates the parent. This means that the init deamon becomes
> the surrogate parent of that child. When pppd exits, it is init that will be
> notified, not the parent such as diald. The -detach option of pppd disables
> this detaching, so that pppd stays with its parent. ]]
>
> Try not giving pppd a chance to do a graceful exit: use ``kill -KILL'' (the
> same as -9) so that diald can't get the impression that the network is down
> before the tty device is closed. If you kill the processes with -TERM, you are
> giving them a chance to perform some activities before they exit. The pppd
> daemon doesn't simply close the terminal device when it exits gracefully; it
> proceeds in steps, by first hanging up by dropping DTR, etc. When it's doing
> this, it's perhaps possible that the diald will launch another pppd. (I don't
> have the source code for diald handy, but I could investigate more deeply, if
> you want).
>
> > I guess another alternative is to somehow automatically rebuild my firewall
> > rules.
>
> Why do your firewall rules have to depend on whether your device is ppp0 or
> ppp1? First of all, does your ``permantent'' connection have a static or
> dynamic IP?
========================================================================
Stephen Davies Consulting
[EMAIL PROTECTED]
Adelaide, South Australia. Voice:
61-8-82728863
Computing & Network solutions. Fax: 61-8-82741015
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
