Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 11 Jan 2002 11:03:49 -0800 (PST)
From:      Julian Elischer <julian@elischer.org>
To:        Colin Henein <freebsd-questions@juicer.orange-carb.org>
Cc:        freebsd-net@freebsd.org
Subject:   Re: pppoe problem (not tcpmssfixup)
Message-ID:  <Pine.BSF.4.21.0201111100160.59070-100000@InterJet.elischer.org>
In-Reply-To: <20020111170554.23144.qmail@mail.orange-carb.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Hard to say, but it does sound like a fragmentation/oversize problem.


try setting the mtu and mru down more
and I don't see proof that the mssfixup is turned on.....

how about you try connect to a site that fails while looking at what is on
the pppoe line with tcpdump (use -i fxp0 or whatever)
it can interpret pppoe and the ppp within it.
maybe it can also do the tcp within that too.
if so you may see what is actually being sent out
which may be of use.


On Fri, 11 Jan 2002, Colin Henein wrote:

> Greetings freebsd-net types.
> 
> I am having a serious problem with pppoe networking on my 4.4-release
> system. I realise that this isn't a list devoted to answering user
> questions, but I have tried freebsd-questions, and I think this issue
> is a bit too technical for that forum.
> 
> I hope you understand that I've tried to solve this problem myself for
> months, and I've thoroughly researched related problems (i.e. black
> hole routers) on the net. I've tried to find relevant answers in the
> archives of this list, and I've tried to contact others on the net to
> no avail.
> 
> I should mention up front that I've been using FreeBSD for many years
> now, and have a solid understanding of IP and related protocols,
> including reasons for and mechanisms used in IP packet
> fragmentation. If things get too hairy I can get my copy of Comer
> out. So if you think you can help, don't hold back the technical
> details on my account...
> 
> I hope you'll indulge this question, as you're really my last resort.
> 
> I am on ADSL service (magma.ca), and the FreeBSD box is connected
> directly to my Nortel 1meg modem. I have an "inside" network as well,
> which pretty much only sends and receives web traffic. I have no ports
> that open through onto the inside network, so all incoming connection
> requests terminate on the FreeBSD box itself.
> 
> I have the classic ADSL problem of not being able to get replies back
> from certain websites (whether telnetting to port 80 from the FreeBSD
> box, or using IE on my powerbook on the inside network). This I can
> deal with through proxies.
> 
> The really critical problem, though, is that I also am running an SMTP
> server (qmail) and have trouble receiving emails from certain sites.
> 
> If I remove the FreeBSD box from the loop and plug my powerbook
> straight into the 1meg modem then I can hit all the websites that are
> problemmatic with FreeBSD. This makes me suspect either my
> configuration of FreeBSD (more likely) or the OS itself (less likely).
> 
> I understand the black hole router problem, but an upgrade to
> release-4.4 with the default tcpmssfixup option on ppp doesn't solve
> the problem. (My ppp.conf file is appended to the end of this
> message.)
> 
> The www problem I can get around with proxies, but the SMTP problems
> are more serious. I am dropping emails from a few hosts on the floor,
> and a few specific mailing list subscribers are having serious
> problems. What happens is that inbound SMTP connections hang after my
> FreeBSD box sends the 'go ahead to send data' response. Here is a
> captured dialog between my mail server (mail.orange-carb.org) and one
> of the problematic senders (tomts9-srv.bellnexxia.net):
>         > 220 mail.orange-carb.org ESMTP^M
>         < EHLO tomts9-srv.bellnexxia.net^M
>         > 250-mail.orange-carb.org^M
>         > 250-PIPELINING^M
>         > 250 8BITMIME^M
>         < MAIL FROM:<normand.levert@sympatico.ca>^M
>         > 250 ok^M
>         < RCPT TO:<members@hotrak.ca>^M
>         > 250 ok^M
>         < DATA^M
>         > 354 go ahead^M
>         > 451 timeout (#4.4.2)^M
>         > [EOF]
> To me this sounds very similar to the black hole router problem, in
> that the connection just sort of times out when bulk of the data is
> coming.          
> 
> As those familiar with Canadian networks might expect, my attempts to
> contact anyone technical at bellnexxia's NOC, or sympatico's IS people
> have been met with resounding silence.
> 
> I desparately need a solution to this that I can implement
> myself. Even my own ISP will be hard to convince since they explicitly
> will not support anything that isn't Mac/Windows/Linux.
> 
> I'm hoping that someone in this forum will be able to shed some light
> on this situation, offer some suggestions for how I can diagnose this
> problem more completely, and perhaps even tell me how I can fix
> things.
> 
> Thanks for considering this issue.
> Colin
> 
> 
> --------- ppp.conf -----------
>         |magma:
>         | set redial 30.120 0
>         | set reconnect 120 9999999
>         | set device PPPoE:ed1
>         | set MRU 1492
>         | set MTU 1492
>         | set authname #########
>         | set authkey #########
>         | set log Phase
>         | set dial
>         | set login
>         | set ifaddr 64.26.169.169 10.0.0.2/0 255.255.255.255
>         | set timeout 0
>         | set cd 5
>         | enable lqr
>         | set lqrperiod 10
>         | set crtscts off
> --------------------------------
> 
> 
> 
> 
> --
>              || when we're little kids maybe we need stories 
> Colin Henein || to help us go to sleep. but sooner or later we 
>              || need stories to help us wake up... -- Utah Phillips
> 
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-net" in the body of the message
> 


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0201111100160.59070-100000>