Reflector 2.40b4 / 2.50b1 probs

Ben Anderson (B.Anderson@lut.ac.uk)
Tue, 20 Dec 1994 05:52:25 -0500


Season's greetings..

This post is all about some problems I have run into with reflectors - if
you don't know about reflectors please don't bother to read further!

Our setup is :- two 2.30 or earlier reflectors (call them L's and B's) are
tied together using UNICAST_REF and have been working fine until now. L's
reflector is also multicast enabled and we can therefore use nv3.2 to check
out who's hanging around on the two reflectors.

I have recently updated L's reflector from 2.3 (version ?) to 2.40b4 and
have noticed some odd behaviour as a result.

- all video streams from B's reflector are now garbled, ie CU-SeeMe shows a
grey window - L's reflector knows video should be arriving and passes this
knowledge on to CU-SeeMe, but CU-SeeMe will not display the video.

- possibly related behaviour: I used nv3.2 to have a look at all the video
being multicast from L's reflector and it attempted to display the 'lost'
video coming via B's reflector. BUT, it claimed wildly varying fps of 30 -
60 and similarly yo-yoing bandwidth usage of 100 - 400 kbs. Just in case
this was not a fictitious stats report I killed everything PDQ. On trying
again I noticed the same behaviour but also noticed that nv appeared to be
trying to display the same video packets several times over as if it were
receiving multiple copies from L's reflector (?). (This is not a problem
with the new reflector's multicast code - video from Macs attached directly
to L's reflector (2.40b4) does not show this odd behaviour.) If this is the
case then presumably CU-SeeMe is refusing to display this garbled video,
hence the grey window.

And so to the questions:

Is version 2.40b4 backward compatible with 2.3? If a 2.40b4 reflector is
tied to a 2.3 reflector (or earlier) should I expect strange behaviour? Is
the problem something to do with differences in the stats reports? Was it
really blasting that much video out?

Finally, I then updated again to 2.50b1. I thought this might solve things.
Silly me. The above problem still exists and in addition, I have noticed
that the MOTD function does not seem to work anymore :-(

If anyone has had similar problems and can suggest solutions I'd be glad to
hear..

cheers
Ben

******************************************************************************
LUTCHI Research Centre, UK.
<a href="http://pipkin.lut.ac.uk/~ben/">My Card</a>.