Bug in multicast code solaris 9 (and 10?)

Question

Hi,
I am working with a client who needs to use
lots (80-100) of multicast addresses.
We are using bge/gld. Here is the problem:
The setsockopt call with the IP_ADD_MEMBERSHIP
argument succeeds even though bge/gld
only has room for 64 multicast addresses.
If I use setsockopt to add 80 group memberships
on different multicast addresses, netstat -g
shows all 80. Sending/selecting any of
the sockets above the 64th has no result.
(Packets are dropped because gld
has failed the group membership add). I found a way to make it work using:
multisize = 100;
in /platform/sun4u/kernel/drv/bge.conf
This allows up to 100 different mcast addresses.
(But this is not documented anywhere as far as
I can tell.)
However, I think the error primitive that gld
sends up when you run out of free mcast entries
should be propagated all the way up to the setsockopt caller. Interesting that the multicast
loopback has no trouble with sockets after 64 (because IP doesn't record the error from gld???).
At any rate, if you need a test, let me know and I'll send it. Hopefully, someone who is working on the
kernel can take a look. I've spent some time and it
looks like the DL_TOOMANY error primitive that gld
sends up from gld_enable_multi() is being dropped
on the floor.
max

Answer 1

In snmpd.conf file add these entries.
rocommunity public
syslocation "System administrators office"
syscontact "System administrator"
sysservices 72
And then start snmpd .Then to do a snmpwalk juz use
~#snmpwalk -c public -v 1 localhost
Assuming that you have compile and configure with SNMP v1 support