Prev: Madwifi doesn't work with my Atheros card
Next: Logging dropped packets via netfilter/iptables
From: Günther Schwarz on 21 Mar 2010 12:45
Marc Haber wrote:

> Günther Schwarz <strap(a)gmx.de> wrote:

>> It might well be that you are talking to the very
>>same implementation on competing 3G networks.
>
> So you're talking equipment on the network side? That might be possible,
> yes.

Ignoring the details completely there has to be some kind of terminal
server on the other end of the 'line' which negotiates tunneling of IP
over the mobile phone network with the mobile phone or USB modem. This
machine will or will not provide a correct DNS address. I would not be
surprised if such a machine was offered by one vendor only.

Günther
From: Moe Trin on 21 Mar 2010 20:21
On Sun, 21 Mar 2010, in the Usenet newsgroup comp.os.linux.networking, in
article <ho54t6$8vj$1(a)news1.tnib.de>, Marc Haber wrote:

ibuprofin(a)painkiller.example.tld.invalid (Moe Trin) wrote:

>> Last I bothered to look at the diff file, it had a change adding
>> 'auth' to the /etc/ppp/options file:

[...]

>> which probably came from a note added to the ChangeLog for ppp-2.3.0
>> in 1997.

>Then one should file a bug in the Debian BTS to have the comment
>removed from the Debian package.

I filed one back in ~1999, but it was never even acknowledged. I had
other things to do, and never followed up.

>Such explanations tend to outlive their purpose since one needs
>really intimate knowledge of the Package to be able to judge whether
>it still applies or not.

One would _hope_ that the package maintainer had such knowledge - and
the information needed is part of the original tarball.

>> WvDial is built on the wrong premise - it expects to be able to log
>> in to a UNIX shell on the remote, and wants to find a Login: prompt
>> to do so. That this mode of operation became rare when microsoft
>> invented the telephone back in 1995 is apparently lost on the
>> authors of wvdial.

>Indeed. Stupid mode works fine though.

Wvdial was publicly released in late 1997 - two years after the use
of RFC1334 was made the default mode for windoze. "Stupid" mode didn't
appear until April 1998, and even then the idiot authors misunderstood
why it was needed. And even in their latest man page, they are so
sure their intelligent dialer is correct that they disable "Stupid"
mode by default. Talk about out of touch with reality.

>It leaves some wishes as well, but they are not half as annoying as
>chat(8)'s shortcomings.

Personally, I found John Hasler's 'p commands' more than enough of a
hand-holder. Most of the features and options in chat(8) and pppd(8)
are never used anyway.

Old guy.
From: Moe Trin on 21 Mar 2010 20:22
On Sun, 21 Mar 2010, in the Usenet newsgroup comp.os.linux.networking, in
article <slrnhqbhs4.1b1.whynot(a)orphan.zombinet>, Eric Pozharski wrote:

>Moe Trin wrote:

>> Some time back in the late 1990s, the Debian maintainer added 'auth'
>> as a required setting in /etc/ppp/options. The only time I've needed
>> the 'noauth' option is when I've been using dialin as a backup to the
>> existing broadband link (which had some reliability problems).

>Probably that's why I really need this ('auth') explicitly unset.

The original item came from a note in the ChangeLog file for version
2.3.0 (see Changes-2.3, about line 430) where it says:

My intention is to make the `auth' option the default in a
future release.

then look at the second paragraph in the 'What was new in ppp-2.3.6.'
section, about line 255. In the mean time, the Debian maintainer
added the 'auth' option to the default /etc/ppp/options, and never
removed it when the change was made in 2.3.6.

>(That seems to be somewhat unfortunate decision for my reality. None
>my ISP ever provided ways to authenticate themselves

I don't think I've ever heard of _ANY_ ISP willing to authenticate
themselves. Simple reason - windoze clients are not configured to do
so, and ISPs do what windoze requires.

>If I have authentication tokens then I just fill 'chap-secrets' *and*
>set 'noauth' anyway.

I set the secrets in /etc/ppp/secrets, and then create links named
/etc/ppp/pap-secrets, /etc/ppp/chap-secrets and /etc/ppp/srp-secrets
all pointing at this file so I don't _care_ what authentication scheme
the ISP wants. PAP, CHAP-MD5, CHAP-MSv1, CHAP-MSv2 or EAP all work.
My /etc/ppp/options file does not have an 'auth' line.

>Mar 21 09:02:56 agentsmith pppd[26007]: Starting link

[...]

>Mar 21 09:03:06 agentsmith pppd[26007]: sent [IPCP ConfReq id=0x8
<addr 178.92.95.146> <ms-dns1 213.179.249.151> <ms-dns2 213.179.249.152>]
>Mar 21 09:03:06 agentsmith pppd[26007]: rcvd [IPCP ConfAck id=0x8
<addr 178.92.95.146> <ms-dns1 213.179.249.151> <ms-dns2 213.179.249.152>]
>Mar 21 09:03:07 agentsmith pppd[26007]: rcvd [IPCP ConfReq id=0x5a]
>Mar 21 09:03:07 agentsmith pppd[26007]: sent [IPCP ConfAck id=0x5a]

Eleven seconds for the link to come up. Keep your eye on this to
gain a feel for the UMTS link wakeup time.

Old guy
First  |  Prev  | 
Pages: 1 2 3 4 5 6 7
Prev: Madwifi doesn't work with my Atheros card
Next: Logging dropped packets via netfilter/iptables