SUMMARY: Network issue in Solaris 10 update 4

From: francisco roque <>
Date: Thu May 01 2008 - 12:19:39 EDT
No resolution to the issue, but we've found some work-arounds, and here 
are some answers to suggestions i've received (thanks for all the 
suggestions!), plus some more info:

- Neither ipmp nor ipf are enabled on the machines with issues.  No other 
firewall is present.

- I get timeouts even when the two machines are in the same subnet.

- I've re-verified that when both the switch and the server are set to 
auto-negotiate, they are appropriately configured the same.  Problems still 

- We've tried hard-setting the switch and a server to 1000/full duplex and 
100/full duplex and still have problems.

- `netstat -i` reports no errors or collisions.

- I am using `ping` for most of my testing, but ssh sessions also time out and 
http connections hang.

- If i ping the 10u4 server while simultaneously pinging anything else *from* 
the 10u4 server, then my ping rates drop to 1% packet loss.

- only changes i've read about to networking are the DAD changes, and
unfortunately the designers decided not to allow us to turn off DAD.

- placing the server behind our pix firewall, running in transparent mode, 
problem persists.

- placing the server behind our pix firewall, running in nat mode, problem 

The production machines in question were eventually going behind the pix, 
so for now we have a workaround - keep them behind the pix and away from 
whatever is downing u4/5 on our network.

Once we get more time we'll try to troubleshoot the issue further to look 
for a real solution - for now project time limits require the workaround 
to be enough.  It really feels like a mac cache issue but it's strange 
that it only affects u4/5 and that u3 in the same physical/logical setup 
is fine.

Thanks for all your help & suggestions,

sunmanagers mailing list
Received on Thu May 1 12:33:03 2008

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:44:10 EST