DL360 NIC Problems

[Adrian Patterson]

We have a ProLiant DL360 running Windows 2000 Server SP4 which has been falling off the network occasionally. It has two HP NC3163 Fast Ethernet NICs, one of which is enabled at any one time, but the problem has been observed with both cards.

I don't know if it's related, but the event log shows the cards going down and back up instantly, i.e. within 1 second (see below). This has happened numerous times, but no events are logged at the point the machine falls off the network.

---------->8----------
Event ID 4 - Adapter HP NC3163 Fast Ethernet NIC: Adapter Link Down
Event ID 18 - Adapter HP NC3163 Fast Ethernet NIC: Adapter Link Up: 100Mbps Full Duplex
Event ID 4201 - The system detected that network adapter HP NC3163 Fast Ethernet NIC was connected to the network, and has initiated normal operation over the network adapter.
---------->8----------

I've upgraded the NIC drivers to the latest version (7.1.12.0) and the machine has not fallen off the network since then (a week ago), but the above events have been logged since then.

Does anyone have any idea why the machine may be falling off the network, and what's causing the events to be logged?
I've found a thread from February indication that there may be a design problem on the motherboards (http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=421266).

The machine's BIOS version is 04/02/01.

Thanks,

Ade.

[Ron_161]

Adrian,

Had similar problem when using auto detect. The network card tries to negotiate a speed and temporarily drops off the network. Try setting your network card to a static speed if you haven't already.

[Chris Turnbull]

Ron's hit the nail on the head - check what speed of switch port your server is going back to and match it, we use 100Mb switches for most of our boxes and have experienced problems in the past with cards left set to auto-neg.

[Mark Maiden]

Adrian,

I am not sure if you have done this, but try plugging both NICs to the same switch?
I to have had a similar problem, but it was the way the network was designed. I had applied all the latest patches and still to no avail. The issue ended up being that there was an extra switch in the loop.
Our design was to have the NICs of the servers, one on each core router (1 Gb switch modules mounted in the core). The problem was that the second core was on another floor and a 1 Gb switch was inserted with a fibre run back to the second core.
This is where the problem arose â another device (switch) in the loop.
Patching the servers to the same core all is okay. We hope to change the network in the not to distant future for resilience.

Hope this hel