Subject: Re: pppd and IPCP failure
To: Brian C. Grayson <bgrayson@marvin.ece.utexas.edu>
From: Frederick Bruckman <fb@enteract.com>
List: netbsd-help
Date: 08/30/1998 08:38:44
On Sun, 30 Aug 1998, Brian C. Grayson wrote:

>   For months, I connect to my ISP only about 50% of my attempts.
> pppd -d with syslogging of local2.* allowed me to fix a few
> things (which brought the connection rate from 25% to its
> current 50%).  But I still notice some weirdness w.r.t. IPCP
> negotiation during the failed attempts:
> 
> pppd[176]: sent [IPCP ConfReq id=0x1 <addr 0.0.0.0> <compress VJ 0f 00>]
> pppd[176]: rcvd [IPCP ConfReq id=0x9 <compress VJ 0f 00> <addr 0.0.0.0>]
> pppd[176]: sent [IPCP ConfRej id=0x9 <addr 0.0.0.0>]
> pppd[176]: sent [IPCP ConfReq id=0x1 <addr 0.0.0.0> <compress VJ 0f 00>]
> pppd[176]: rcvd [IPCP ConfReq id=0xa <compress VJ 0f 00> <addr 0.0.0.0>]
> pppd[176]: sent [IPCP ConfRej id=0xa <addr 0.0.0.0>]
> 
>   This continues for 30 seconds, at which point my machine gives up.
> 
>   So, why does my machine ask for <addr> <compress>, and then
> reject <compress> <addr>?  Aren't requests 0x1 and 0x9
> identical as far as their meaning?  Or is the remote end
> supposed to have filled in the 0.0.0.0 in request 0x9 (and 0xa and ...)?
> 
>   My options file is appended below.  Thanks for any help -- I
> tried looking at the source code, but quickly got lost.
> 
>   Brian
> -- 
> /dev/tty02
> 115200
> defaultroute
> crtscts
> lock
> modem
> ##  These two end up trying <compress VJ 0f 00>
> vj-max-slots 16
> novjccomp
> mtu 572
> mru 572
> disconnect '/etc/ppp/ip-down'
> connect '/etc/ppp/redialchat'

Why do you have "novjccomp"? What happens when you omit that? Methinks 
you also need to have "noauth" in there somewhere, unless you have a wildcard
in /etc/ppp/pap-secrets.

Also, "mtu 572" is somewhat weird. IIRC the acceptable minimum is 576.
Another thing--"ip-down" is called automatically when your connection
drops. Making it your disconnect script means it will be called twice.

Here's part of my /etc/ppp/peers/enteract

noauth                  # Don't require host to authenticate.
nobsdcomp
nodeflate
vj-max-slots    16
mtu 576
207.229.133.236:        # Enteract will be phasing out static id's for dial-ups,
ipcp-accept-local       #      unfortunately.
defaultroute            # You need this!

And here's what you're supposed to get for the vj part. I've never omitted
the local ip address, but I think that's OK if you don't have a static ip.

Aug 14 11:52:34 fb pppd[6367]: sent [IPCP ConfReq id=0x1 <addr 207.229.133.236> <compress VJ 0f 01>]
Aug 14 11:52:34 fb pppd[6367]: rcvd [IPCP ConfReq id=0x1 <compress VJ 0f 00> <addr 207.229.129.246>]
Aug 14 11:52:34 fb pppd[6367]: sent [IPCP ConfAck id=0x1 <compress VJ 0f 00> <addr 207.229.129.246>]
Aug 14 11:52:35 fb pppd[6367]: rcvd [IPCP ConfAck id=0x1 <addr 207.229.133.236> <compress VJ 0f 01>]
Aug 14 11:52:35 fb pppd[6367]: local  IP address 207.229.133.236
Aug 14 11:52:35 fb pppd[6367]: local  IP address 207.229.133.236
Aug 14 11:52:35 fb pppd[6367]: remote IP address 207.229.129.246
Aug 14 11:52:35 fb pppd[6367]: remote IP address 207.229.129.246

Caveat emptor: I have episodic problems connecting to my isp as well. 
Most of that seems to be the modem. It freezes up solid as soon as the isp
answers, and I have to cycle power on it.