Date: Wed, 28 Jun 1995 14:49:02 -0700 From: Amancio Hasty <hasty@rah.star-gate.com> To: Scott Mace <smace@crash.ops.neosoft.com> Cc: hackers@freebsd.org Subject: Re: Space Shuttle Mission is being broadcasted Message-ID: <199506282149.OAA01416@rah.star-gate.com> In-Reply-To: Your message of "Wed, 28 Jun 1995 15:33:08 CDT." <199506282033.PAA07529@crash.ops.neosoft.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>>> Scott Mace said:
> >
> >
> > Howdy,
> >
> >
> > If you have an mbone connection the Space Shuttle Mission is being
> > broadcasted.
> >
>
> Yes!, I'm getting good video right now... audio is choppy but thats because
of my internal config...
>
> getting between 1-5fps at 130kbps
>
> Scott
Well, there has been some slight problems on the MBONE.
BTW: Does anyone if we can generate the bogus ICMP messages mentioned
in the enclosed mail.
Enjoy,
Amancio
Return-Path: geoffw@nexsys.net
Received: from nexsys.nexsys.net (nexsys.nexsys.net [192.188.198.1]) by rah.star-gate.com (8.6.11/8.6.9) with ESMTP id KAA00942 for <hasty@star-gate.com>; Wed, 28 Jun 1995 10:59:28 -0700
Received: by nexsys.nexsys.net (8.6.10/SM-8.6.4)
id LAA29248; Wed, 28 Jun 1995 11:01:47 -0700
Date: Wed, 28 Jun 1995 11:01:47 -0700
From: geoffw@nexsys.net (Geoff White)
Message-Id: <199506281801.LAA29248@nexsys.nexsys.net>
To: hasty@star-gate.com
Subject: Re: host viviane.usl.edu still trashing shuttle video stream
Cc: ramsey@netcom.com
Hmmm...
----- Begin Included Message -----
>From list-mgr@ISI.EDU Wed Jun 28 09:57:24 1995
Return-Path: <list-mgr@ISI.EDU>
To: Van Jacobson <van@ee.lbl.gov>
Cc: mbone@ISI.EDU, mahler@usl.edu
Subject: Re: host viviane.usl.edu still trashing shuttle video stream
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Id: <26459.804354074.1@msf.psi.net>
Date: Wed, 28 Jun 1995 11:41:14 -0400
From: "Mark S. Fedor" <fedor@msf.psi.net>
Content-Length: 2981
X-Lines: 57
Status: RO
FYI... I've been running RTP traces on the shuttle video at mae-bone.psi.net
(on mae-east) all morning and I have an average loss rate of 30%.
At this time, the loss rate continues at 30%.
Mark
-------
> As Bill Fenner pointed out yesterday, host viviane.usl.edu
> (130.70.40.162) appears to be causing the >50% packet loss that
> most people are seeing on the Shuttle video session. It is
> causing the loss by sending an ICMP unreachable packet in
> reponse to every multicast packet put on the 130.70.40 subnet at
> USL. For a few hours yesterday evening, it appears that
> alpha.noc.usl.edu (the mbone tunnel endpoint at USL) was off the
> mbone and the NASA video reception here was perfect (0% loss).
> At around 12:50am, I started to see ICMP unreachables from
> viviane again (sent in response to my session msgs in the
> shuttle audio session) and the NASA video loss rate immediately
> went up to 50%.
>
> There are already ~400 people tuned into the Shuttle mission
> and there will probably be a couple thousand trying to watch
> the MIR docking. It would be a shame if this one broken host
> screwed up the video for people all over the world. It would be
> best if viviane.usl.edu could be moved to a subnet that didn't
> get any multicast traffic or just powered off (the machines that
> normally generate these bogus ICMPs are IBM RTs & Unisys machines --
> often these are just as useful with the power off as on). If
> it can't be powered off or moved, perhaps a filter can be installed
> on the local gateway to discard all ICMPs from viviane. If USL
> can't do anything, perhaps Georgia Tech could temporarily disconnect
> the tunnel from houdini-fddi.gatech.edu to alpha.noc.usl.edu until
> USL has time to fix the problem.
>
> For the curious, the reason that viviane's ICMPs are trashing
> the video stream is that the kernel interprets ICMP unreachables
> as an error (they probably would be an error if the traffic were
> unicast) which causes the next send that nv does to be aborted
> with an ENETUNREACH error. Nv effectively ignores the error but
> it does cause the packet that it was trying to send to be
> discarded. If the nv packets are well spaced, the result is
> that every other packet gets discarded (50% loss). (The reason
> that vat audio is working is because vat looks for EUNREACH
> errors and resends the packet if it gets one.) Viviane is
> probably not running the multicast apps or a multicast kernel --
> it is probably just some host with a broken IP stack that is
> mis-handling any multicast traffic that happens to appear on the
> local wire. The reason it's cobbering the Shuttle session is
> because Steve Mahler is tuned into the Shuttle sessions on host
> alpha.noc.usl.edu (which is not misbehaving) and that causes the
> shuttle multicast traffic to get put on the 130.70.40 subnet
> where viviane lives. Until viviane gets fixed, any video
> session that Steve tunes into will get trashed.
>
> - Van
----- End Included Message -----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199506282149.OAA01416>