Re: Fwd: SSM Video Group
At 11:30 -0500 12/14/04, Marshall Eubanks wrote:
>There is a persistent, off and on again, problem between Level 3 and I2 which
>affects all multicasts into I2 from Multicast Tech. The symptom is
>that multicast
>packets only flow for about 5-10 seconds, and then stop for alost 5
>minutes, then repeat the
>pattern.
>
>As part of the debugging for this I updated the beacon server here
>to 1.1. I notice
>that the beacon loss stats do NOT seem to match this. For example,
>the problem occurs between
>here and IU, but the loss right now between
>63.105.122.14 and wingrid.ads.iu.edu (which I presume is Matt's
>beacon) is currently listed at
>0%.
>
>How can this be ? This sounds like a bug to me.
I know that the 1.1-0 server that I'm running (on solaris) does not
correctly display the data, it doesn't even always show all the hosts
that are transmitting. the TCP connections don't connect right, I
can watch the error messages scroll by on the server. when hosts do
come up, their data is updated sporadically, but it still displays.
Mitch is aware of the problem, and I'm pretty sure he found some
linux issues as well (which is the platform of the dast.nlanr beacon
server).
for me the local_loss page which is made based on data in the
multicast (UDP) packets, seems to be more accurate. A quick look at
the local_loss page on beacon.dast.nlanr.net shows a vastly different
view of the "state of multicast". I wouldn't go so far as to say
that it is right (I don't _think_ the connectivity is that broken...)
but IMHO there shouldn't be that huge a difference.
I know that there should be a 1.2 version soon, and that it will be
addressing some of these problems, but I don't know how much of this
is addressed or for certain when it will be out. Mitch will need to
comment more on that, but I know he's been really busy lately.
>
>A long discussion of the problem is below.
>
>Regards
>Marshall Eubanks
--
-debbie
Debbie Fligor, n9dn Network Engineer, CITES, Univ. of Il
email: fligor@uiuc.edu <http://www.uiuc.edu/ph/www/fligor>
"Every keystroke can be monitored. And the computers never forget."