All things related to multicast

["Craig M. Votava" <>]

Folks-

I'm seeing something strange here, and I'm hoping for some
help. This is occuring behind the Lucent firewall, on our
private multicast network.

SYMPTOMS:
1.) Start sdr on any machine
2.) Sdr runs fine for some number of minutes, usually not
        greater than five.
3.) Sdr dies, no error message, no reason given.

DEBUGGING INFO:
1.) Ran sdrlisten.pl on the sdr address/port (224.2.127.255/9876)
        and saw the following excerpt:
=====================================================================
v=0
o=ccharper 3208608641 3208608910 IN IP4 135.185.57.30
s=L-Channel Test
i=Multicast test from the L-Channel Lisle/Naperville facility
u=http://llsalesandmarketing.web.lucent.com./lchannel.html
e=L-Channel Lisle/Naperville 
<>
p=L-Channel Lisle/Naperville 630.224.2674
c=IN IP4 224.2.130.65/63
t=3208597200 3208964400
r=1d 6h 0
a=tool:sdr v2.4a6s3
a=type:test
m=audio 30838 RTP/AVP 0
c=IN IP4 224.2.130.65/63
a=ptime:40
m=video 59286 RTP/AVP 31
c=IN IP4 224.2.134.98/63
!a\167\007\064\006\161x
\034\055\026\137n\1428^T
\106\157\021\166^]
<More lines of binary stuff>
=======================================================================
        Further testing confirmed that as soon as the binary stuff
        showed up, sdr died.

2.) Ran netsnoop (SGI network sniffer program) on my network segment
        for any traffic on 224.2.127.255. Here's what I saw:
=======================================================================
0000:   len   60   time 09:34:53.156796
    ether:  src nimrod.ih.lucent.com      dst 1:0:5e:2:7f:fe/IP-mcast  
            type ip    
    ip:     v 4       hl 5      tos 0     len 28    id 19471  off 0
            ttl 1     p igmp    sum 0x9208
            src nimrod.ih.lucent.com      dst 224.2.127.254            
    igmp:   version 1           type 6              unused 0   
            cksum 0x89fe        group 224.2.127.254  

0001:   len   60   time 09:35:11.416994
    ether:  src nimrod.ih.lucent.com      dst 1:0:5e:2:7f:fe/IP-mcast  
            type ip    
    ip:     v 4       hl 5      tos 0     len 28    id 20689  off 0
            ttl 1     p igmp    sum 0x8d46
            src nimrod.ih.lucent.com      dst 224.2.127.254            
    igmp:   version 1           type 7              unused 0   
            cksum 0x88fe        group 224.2.127.254  

0002:   len   60   time 09:35:11.417415
    ether:  src 2:e0:52:e9:ba:d0          dst 1:0:5e:0:0:1/IP-mcast    
            type ip    
    ip:     v 4       hl 5      tos 0     len 28    id 1      off 0
            ttl 1     p igmp    sum 0xda5b
            src il106171-rdi-l2s1-9e517.i dst 224.2.127.254            
    igmp:   version 1           type QUERY          unused 0x64
            cksum 0x8e9a        group 224.2.127.254  
=======================================================================
        The first 2 packets are the joins that my machine (nimrod) put out
        on the network because of running sdr. The third packet is the one
        that causes sdr to die. The source is a foundry L2 switch.

3.) Further sniffing with different sniffing tools shows the following:
=======================================================================
"IGMP"," Type 6, Ver2 Membership Report"
"IGMP"," Type 6, Ver2 Membership Report"
"IGMP"," Type 7, Leave Group"
"IGMP"," Type 1, Host Membership Query"
=======================================================================

I'm confused. It looks like the foundry L2 switch is sending a Host
Membership Query to the sdr address/port, and my protocol stack is
sending it to the sdr program, which exits. Is this supposed to happen?
Do I have a bad version of sdr (I've upgraded to the latest one
I could find)? What's going on here!?!

Any help is appreciated!

Thanks

-Craig Votava
Lucent Technologies
Lisle IL USA

PS - FYI: We had a total of 380 desktops connected to our internal
multicast feed of the disaster coverage.