SUMMARY: Problem with MAC addreses

Hi,

	Thanks for your help. I was told to turn out Gigabit capabilities,
set ndd parameters to force 100fdx and disable auto-negotiation, to
install patch 111883, and to disable auto-everything in the switch if a
Cisco switch is used.

	The solution was to force 100hdx (half duplex) in the Cisco
switch. Of course half duplex is not the best setting but the switch is
not under my control. The actual setting works so it is enought for me.

	I did not install the patch which is for Solaris8 and my system
is Solaris9 

Best Regards,
	Jordi Vidal


---------- Forwarded message ----------
Date: Fri, 11 Jul 2003 17:36:13 +0200 (CEST)
From: Jordi Vidal <jordivi@wtransnet.net>
To: sunmanagers@sunmanagers.org
Subject: Problem with MAC addreses

HI all,

	I have a extrange network problem with a Sun Fire V480. It has a
dual network adapter ce0 and ce1 in a PCI card. They work at 100Mbps half
duplex. I have orca monitoring the hardware performance. Orca plots severe
network problems on its graphs and the underlaying "se" tools
(live_test.se) complaints about network problems: "Check hardware or
cables". FTP transfers are very slow, about 1Mbps or less.

	The setup was as follows:
ce0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
        inet 192.168.250.4 netmask ffffff00 broadcast 192.168.250.255
        ether 0:3:ba:2c:9a:34

Checked cables and switches, I did some tests and found that the NIC works
fine enabling the second interface and messing up with MAC addresses. 
Changing primary NIC MAC address works ok. 

ce0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
        inet 192.168.250.4 netmask ffffff00 broadcast 192.168.250.255
        ether 0:3:ba:2c:9a:35
ce1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 11
        inet 192.168.250.6 netmask ffffff00 broadcast 192.168.250.255
        ether 0:3:ba:2c:9a:34

With this, FTP's to ce0 works ok. Not to ce1 (both are in the same LAN).  
In my orca graphs the plot of "network problems" reduced at half but not
disappear. I observed also in the plots that packets come in by ce0 and go
out by ce1. I tried to shutdown ce1, then "hardware" network problems 
appear again. I put it up again but there are still problems. Messing
with the MAC address finally I can put the network in the same state (ftps 
to ce0 ok, not to ce1, asimetric traffic) It seems to be some "memory" 
here.

	Does this make sense for someone? Have I a broken NIC?

NOTE: No other network errors presents (no collisions, no nocanputs, 
nodeferreds ...)

Regards
	Jordi Vidal
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers