Intel Pro/1000 PT

[acole1]

I am working with an ASRock 4CoreDual-VSTA motherboard, OS is Server 2003.

It has an integrated 100Mb nic, and I needed a Gb nic so I put a PCIe x1 Intel Pro/1000 PT (EXPI9400PT) in the PCIe x16 slot.

It detects the card, drivers install fine, and it works perfect, until I try and use it.

When I plug it into the network (even using the same cable that works in the onboard nic) it sits for a while looking for a DCHP address, and then comes up with the "Limited or no connectivity" error.

But it doesn't stop there.

After showing that message for a few seconds, the "limited or no connectivity" turns into "a network cable is unplugged" for a seccond, then switches back to "limited or no connectivity," waits for a seccond, the pc "locks up" (I can't do anything with the mouse or keyboard for a couple seconds), and then it goes back to normal (displaying "limited or no connectivity"), and starts the process over.

If I unplug the cable, the process stops and there is no locking or anything strange that happens, even when using the onboard nic.

Any ideas?


Edit:
I ran the diagnostics, and it fails the hardware test on the last item, "Loopback test."

 

[nweaver]

unplug the cable, does the loopback still fail?

It sounds like you are getting a L2 problem, like a duplex mismatch. is the switch a gig switch, is it hard set to 1000/full, does it show any port errors?

 

[acole1]

Originally posted by: nweaver
unplug the cable, does the loopback still fail?

It sounds like you are getting a L2 problem, like a duplex mismatch. is the switch a gig switch, is it hard set to 1000/full, does it show any port errors?

Unplugged cable: Loopback still fails

The switch is a gig switch: D-Link DGS-1005D

Link speed: auto config



ERRORS:

Sys log event corresponding to the mouse locking up.

Event ID:12
Source: i8042prt
Type: Information
Description: The ring buffer that stores incoming mouse data has overflowed (buffer size is configurable via the PS/2 mouse properties in device manager).



Sys log events that correspond to the nic events.

Event ID: 32
Source: e1express
Type: Information
Description: "Intel(R) PRO/1000 PT Server Adapter Link has been established: 1000Mbps."


Event ID: 4201
Source: Tcpip
Type: Information
Description: "The system detected that network adapter Intel(R) PRO/1000 PT Server Adapter was connected to the network, and has initiated normal operation over the network adapter.

Event ID: 27
Source: e1express
Type: Warning
Description: "Intel(R) PRO/1000 PT Server Adapter Link has been disconnected."


And it just repeats that cycle.

 

[nweaver]

try another port on the switch, and another (premade, not homemade) cable. I'm not sure if the loopback required a special cable or not with these, but that might indicate a hardware failure on the card.

 

[acole1]

Originally posted by: nweaver
try another port on the switch, and another (premade, not homemade) cable. I'm not sure if the loopback required a special cable or not with these, but that might indicate a hardware failure on the card.

Another port: same problem

Premade cable: no change

10mb hub (linked to gb switch), static IP, premade cable, nic forced 10mb link (full or 1\2 duplex): no change


I am beginning to think the card is the problem. I'll see if I can find another PC to test it in.


Edit:
When I use the 10mb hub the errors and lockups stop, but when I leave the settings the same (static IP, link forced to 10mb full duplex) and put it back on the gb switch, the errors start again, which is more evidence towards a L2 problem.

 

[nweaver]

the connect/disconnect cycle is what's bad. Try forcing 100/full on both ends to see if that makes it stable

 

[acole1]

Originally posted by: nweaver
the connect/disconnect cycle is what's bad. Try forcing 100/full on both ends to see if that makes it stable

I can force 100mb full on the nic, but the switch is out of my control. It's just a simple D-Link switch, and I can't think of any others around that I could hard set 100/full.

Would it be worthwhile to test it on another switch (3com Super Stack 3, 3250) that is auto neg 100mb?

I also might be able to try a different gb switch too, if you think that would help anything.


Edit:
I just tested it on another machine, put it in the pcie x1 port (instead of a x16 like in the asrock board) and now it works perfect! I guess it was something to do with the x16 slot.