hello
We have a multicast address (234.234.234.234) that is
configured on each of our 10 proxy servers so that ICP
sibling proxies can efficiently broadcast ICP requests
to each other.
Some percentage of ICP requests (10 to 20%) time out
without ever getting a reply from siblings. This
results in delays and retries for all proxies.
Unfortunately the ICP protocol does not provide much
useful diagnostic information, so the multicast
problem can best be seen by examining ping replies.
It appears that about 20% of multicast requests are
mis-routed (seem to be routed through the wrong
adapter) and lost.
When we ping 234.234.234.234, we should get normally 9
ICMP replies from the remaining proxies. But if the
same ping is allowed to run for a longer time, then it
becomes clear that some multicast traffic is
apparently sent across the wrong network adapters
(because replies come from unexpected addresses).
Also the amplification factor drops from 9 to 7.11
which implies that ICP sibling proxies didn't receive
the multicast requests and thus didn't reply to them.
For instance, we see some ping failures from subnets
other than the ICP subnet. We also miss the 9
multicast replies from the sibling.
Can you please shed some lights on this issue?
thanks,
Mel
___________________________________________________________
To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre. http://uk.security.yahoo.com
Received on Tue Sep 20 2005 - 01:50:29 MDT
This archive was generated by hypermail pre-2.1.9 : Sat Oct 01 2005 - 12:00:03 MDT