*BSD News Article 99530


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!lucy.cc.swin.edu.au!news.rmit.EDU.AU!news.unimelb.edu.au!munnari.OZ.AU!news.Hawaii.Edu!news.caldera.com!enews.sgi.com!newshub1.home.com!news.home.com!newsfeed.direct.ca!news.he.net!dimensional.com!flatland.dimensional.com!not-for-mail
From: mfuhr@dimensional.com (Michael Fuhr)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: PPP allows ICMP, drops TCP & UDP (not ipfw prob)
Date: 12 Jul 1997 21:43:52 -0600
Organization: Dimensional Communications
Lines: 33
Message-ID: <5q9ito$853@flatland.dimensional.com>
References: <5q9a6t$5lv@flatland.dimensional.com>
NNTP-Posting-Host: flatland.dimensional.com
X-Newsreader: NN version 6.5.1 (NOV)
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:44334

mfuhr@dimensional.com (Michael Fuhr) writes:

>     * TCP from me to ISP fails:  tcpdump on my side shows RST going
>       back to ISP (nothing listening on that port); tcpdump at ISP
>       never sees the RST, and ISP continues sending SYN segments.

I forgot to mention that all TCP segments I send are dropped.  If
I initiate a connection, the ISP never sees the SYN segments from my
machine.  Tcpdump at my end shows them on the interface, but tcpdump
at the ISP shows nothing.

Also, attempts to bring up PPP with this ISP often fail with the error
"IPCP: timeout sending Config-Requests".  Turning on pppd debugging
shows the following messages repeating until pppd gives up:

    rcvd [IPCP ConfReq id=0x3 <addr 204.227.16.65>]
    sent [IPCP ConfAck id=0x3 <addr 204.227.16.65>]
    sent [IPCP ConfReq id=0x1 <addr 0.0.0.0>]
    rcvd [IPCP ConfReq id=0x4 <addr 204.227.16.65>]
    sent [IPCP ConfAck id=0x4 <addr 204.227.16.65>]
    sent [IPCP ConfReq id=0x1 <addr 0.0.0.0>]
    ...

I never see this behavior with another ISP, yet this one says his other
users don't have any problems (I know the ISP and have no reason to
doubt him).

I'm trying various pppd options to see if I can find something that
works, but I'd appreciate any suggestions as to what might be wrong.

-- 
Michael Fuhr
http://www.dimensional.com/~mfuhr/