Network cards stops responding

[Anil Persaud_1]

Hello,

we recently upgraded to a HP Proliant ML350 G3 server with the following specifications:
- 2 GB RAM
- HP NC7760 Gigabit Server Adapter
- Redhat Advanced Server 2.1 (2.4.9-e.3smp kernel)
- Oracle9i R2 (9.2.0.1.0)

whenever there appears to be heavy user transactions going on,the network card just freezes up. there are no error messages in the log files. the only solution we have now is to bounce the network interface.

can anyone please help me with this?

[Olivier Drouin]

what kind of cards ?
what driver do you use ?

[Anil Persaud_1]

we use the following driver:
Broadcom Gigabit Ethernet Driver bcm5700 with Broadcom NIC Extension (NICE) ver. 2.0.28 (11/05/01)

we use the following NIC:
Compaq NC7770 Gigabit Server Adapter

[Matt Palmer_2]

Hi,

I had a similar problem, and it turned out to be a problem with the smp kernel I had installed. Is there a chance you can run for a time on just a single cpu kernel, load the network and then see if the problem still presents itself??

regards

Matt

[Anil Persaud_1]

thanks for your suggestion Matt.

we will boot a single cpu kernel, load the network and see how things develop.

thanks again.

[Anil Persaud_1]

Matt,

we booted a single cpu kernel and had no problem for a full day. before we had an average of 3-4 interruptions per day.

on the second day however the same thing happened: the network card stopped responding. after a couple of minutes however it was responding again without any user intervention.

anymore suggestions?

[Matt Palmer_2]

hi,

do you have access to the tcpdump program??

it might be worth running this at intervals, and seeing if you can capture the events leading up to the machine seizing.

Are you running gigabit network or is it just a gbit capable card running 100Base??

If so, it might be an idea to ask HP if that would be likely to cause a problem with the NC7760 and RedHat. Other than that I am afraid I am out of ideas, sorry.

[Stan Halter]

Anil,

Did you ever find a fix to this issue? It seems that you are not the only one with this issue.

I too have this problem where the NC7760 stops responding upon load. The only resolution is to reboot the device. However, the difference here is that we are using Win2k, not Linux.

[Anil Persaud_1]

Stan,

we did 'fix' the problem with the NIC. It appears that RHAS2.1 ships with an old driver which was causing the NIC to become unresponsive after a while. So we downloaded a newer version of the driver from the broadcom site, uninstalled the original version (v2.x) one and installed the new version(v7.x). Since then we have had no more problem with the NIC on the servers running Linux.

We also have a HP Proliant ML350 G3 running win2k. No problem here. The support CD which comes with the server contains the win2k driver for the NC7760.

Hope this helps you.