On Sunday, September 26, 1999 8:18 PM, Rajiv Ghai [SMTP:[EMAIL PROTECTED]]
wrote:
Rajiv:
...
> To connect to my ISP I use the command ./ifup-ppp ifcfg-ppp0 &
> The Linux box dials out, chat script executes and I get connected
> with
> remote and local IP address allocated dynamically. I then
deliberately
> do
> nothing until the connection times out and disconnects after a few
> minutes.
> At the prompt I then telnet any site on the internet and the dialer
> fires
> up again, makes the connection again to my ISP. So demand dialing
> works fine.
> My question is: Is this the normal way that demand dialing is
supposed
> to
> work ?
Yes it is. You can't have a daemon monitor something if it isn't
running.
> Why cant I just telnet at the prompt to fire up the connection the
> first
> time ?
Because of that.
> The second option I mentioned is that I activate the PPP interface at
> boot
> time (from selecting the relevant ppp settings in linuxconf). As the
> linux
> box boots up the machine dials out and connects to the internet. I
> wait a
> few minutes until the connection times out and try a telnet. It fires
> up
> again.
You realise that basically this is the same process as above, right? It
only diverges on the time when pppd is launched.
> It seems to me that for ppp demand dialing to work an initial
> connection
> must be made to the ISP (or other ppp link) manually after booting
> the
> system or automatically at bootup. Is this correct ?
Correct in essence.
> There is no LAN involved here. Just a standalone linux box at this
> time.
> What I did notice in both cases is that after doing a dialout and
> timeout
> the pppd daemon is still running when I check the process list using
> ps ax.
This is normal for daemon processes.
> Is there a way to get the pppd daemon running in demand dialing mode
> at
> boot up without actually dialing out ?
I don't think so, but I can be wrong.
> Has anyone else run into the same problem ?. I saw on the diald
> mailing
> list that several people had this problem using diald but no one
> posted a
> solution. Perhaps they just leave the machine on 24hrs a day so it
> doesnt
> really matter wasting a few minutes of access time during bootup.
This is basically different. diald fires pppd on start-up because it
was asked to. You _can_ avoid this by simply either blocking all DNS
calls and/or not running named, but then you would never get a
connection, unless you use quad-dot notation.
Some have reported that if you start diald _after_ named it won't fire
a connection, but it doesn't work for me.
With diald you _can_ force a link down, but before you must let it go
up or it will retry until a successful up: issue "kill -SIGINT $(pidof
diald)" and it will go down fast without killing diald.
By the way, I don't know why are you so worried with this, be inventive
and take advantage of it. It will only happen once every time you turn
on your box, so it isn't that bad. If the problem is a long time to
timeout maybe you should shorten your timeout options. If your phone
company bills you by impulse check the impulse option for diald. I use
it myself and I take advantage of that connection I'm forced to on
start-up to get and send mail and synchronise the system watch and this
with the cost of only one impulse.
Just one more thing, if you are using diald you won't need the demand
option for pppd, it's redundant. On the other hand it's mandatory for
your intents if pppd is running standalone.
Hope this helps.
---
Lazarus Long
<[EMAIL PROTECTED]>
PGP:
<http://wwwkeys.pgp.net:11371/pks/lookup?op=index&search=0x5C1DC205>
ICQ#: 30062012
-
To unsubscribe from this list: send the line "unsubscribe linux-ppp" in
the body of a message to [EMAIL PROTECTED]
