1650 Ethernet Issue

Dave Johnson dave at wangtrading.com
Tue Oct 22 17:44:01 CDT 2002


Brent--

We also thought e1000 was a superior NIC--until recently.  In the last
week we have run into two big e1000 multicast showstoppers that do not
bode well with our business requirements:

Problem #1: Every hour (or there about) the machine will stall for a few
seconds.

Problem #2: 2.4.18-17smp kernel from RH will lose ~30% multicast packets
with MGEN (20,000 pps at GigE full duplex), in a more "real world"
environment ~5% of our test packets fail.  2.4.18-5smp didn't lose any
packets, but then this goes back to problem #1.

-----

With low traffic going through the servers e1000 nics had little
problems.

-----

Thus for us it doesn't appear to be a temperature thing, but more kernel
related.  FYI--We are targeting mainstream kernel releases at this
point.

- Dave

--------------


Brent DiNicola brentd at cicada-semi.com 
Fri Oct 18 08:50:00 2002 
------------------------------------------------------------------------
--------

In my experience this is a trouble that I have had
with the E1000 cards in both 2500 and P530. They occasionally
die and have to be "reloaded" to work again.. until they do
this again. This might be software but I have found nothing that
fixed it. I now only use the 3com bcm5700 cards.. which work
great and have yet to fail. The E1000 cards, I have used 6 ..
all 6 have done the same thing to varying degrees and 2 have
actually died completely. On this list though.. it appears
people prefer to use the E1000 and have good luck so make sure
to get the latest software information.

Brent

-----Original Message-----
From: linux-poweredge-admin at dell.com
[mailto:linux-poweredge-admin at dell.com]On Behalf Of Hoy, Mark
Sent: Thursday, October 17, 2002 12:20 PM
To: Linux-PowerEdge at dell.com
Subject: 1650 Ethernet Issue


I've a test pgm that blasts data at variable rates out an Ethernet.

I've used various Intel e1000 drivers. Currently I'm testing with
E1000 V 4.1.7 and a 2.4.19 kernel. There is a single 1.4G PIII with 1G
memory.

When I continuously run at 250 Mbps the line, the Ethernet under test
will
eventually fail. I can't tell if this is a h/w or s/w problem....

I've a cron job that takes the interfaces down, removes
the e1000 module, re-installs the e1000 module, and brings the interface
back up. This will fix the problem for a while.

I've noticed on the linux-kernel a thread on the e1000 NICs overheating.
http://www.uwsg.indiana.edu/hypermail/linux/kernel/0210.0/1864.html

Is there a way of getting an extra fan blowing on the chip to see if
it's
heat related? (ie, is there an extra power connector somewhere....)

I've also got a 1650 receiving this stream, it never has a problem.
=-=-
Mark Hoy
Securify Inc.

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge at dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list
archives at http://lists.us.dell.com/htdig/






More information about the Linux-PowerEdge mailing list