HP NC510C losing connection

[MatthewQ]

Yesterday, I installed three HP NC510C 10GBe NICs in three separate non-HP servers. All three servers have 16 CPUs, 64GB RAM and are running 2003 Enterprise x64. The servers are Dell PowerEdge R900s, each with four Broadcom NICs onboard.

Upon installing the NICs, Windows found a driver and installed it. I was able to configure the NICs normally and the NIC functioned fine...

.. for a short time... before the 'cable disconnected' icon appeared in the systray. The NIC's icon in the Network Connection control panel also indicates it is disconnected. Physically looking at the server, the NIC has a blinking link/activity light, and the port on the Cisco 4900M shows an active connection.

If the Network Connection is disabled then reenabled, the status goes to normal, and the NIC will communicate for a short time, before the event is repeated.

I noticed an event from NXP2NIC in the system event log which stated the driver and firmware do not match, update the driver. I updated the driver to 3.4.336.0.

The same situation occurs. After a few minutes of normal functionality, the NIC indicates the cable is disconnected, when it is not, and all physical link indicators are normal.

I'm stumped on this. Any input?

[André Beck]

Hi Matthew,

when Windows automatically assigns a driver to a NIC that is in fact more than just re-branded by HP, it might do so based on assumptions about the chips on the NIC that might not be 100% correct. Have a look whether HP provides a driver download for this NICs, and compare the driver in this package to what Windows uses. If it's different, I'd try that driver. Dunno how well it will run on another vendors platform, but I'd expect it to do just fine.

BTW, any "errdisable" state changes on the interfaces at the Cisco side?

HTH,
Andre.

[MatthewQ]

Hi, I did download a driver from HP's site, and installed it, but the problem persists.

[KarloChacon]

hi

just in case try updating Firmware too

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=329290&prodSeriesId=3244088&swItem=MTX-616f37cd472c4401a6e4ebbd7b&prodNameId=3244089&swEnvOID=1005&swLang=8&taskId=135&mode=4&idx=2

regards

[MatthewQ]

Yeah, no joy there. Trying to do anything with the NXFLASH utility, it can't communicate with the NIC.

It returns the following:

NetUserGetInfo:failed ,

Any switch with that command returns the same.

I'm thinking there's an incompatibility somewhere, or the driver just doesn't work.

[MatthewQ]

I found the problem. I had contacted NetXen via email about the situation, and their tech called me yesterday.

The problem is the driver doesn't function properly with >32GB RAM in the system. The three servers I'm having issues with have 64GB RAM.

[William Gilroy]

I am having the same issue DL585g5 with 8gig of memory.

[Matt Moreno]

I am having this exact same problem with a NC510F. HP has released 3 revisions of the driver for the NetXen NX2 all with the exact same version number, only the date has changed.
My server is a HP DL380-G5 with 16GB of memory.
Is there a driver that we can get from NetXen instead of HP? NetXen's website no longer exists after they were bought out by Qlogic.

[Allen Buswell]

I am having a similar problem on a NC522SFP adapter. It is installed in an HP DL580G5 with 8Gb RAM running Windows Server 2003 Standard x64 Edition on it.
once or twice a week, i get an entry in the system event log saying that "the network link is down. Check to make sure the network cable is properly connected" then one second later, the link comes back. In the teaming statistics, the CRC count has gone up and the Overruns count has gone up on one of the two adapter ports.... the other adapter port remains clean with zero errors. On the physical layer, the Cisco 6509 hasnt noticed a thing.... no errors of any kind, no disconnect..... nothing. and if I bring up the gui showing the NIC statistics on the Windows server, it says its been connected several days.... in other words, it didnt physically disconnect either. The teaming is 802.3ad LACP on both the server and the Cisco 6509. In between network disconnects, the adapter port will handle as much as 30-40 Terabytes of traffic being sent to it. The traffic is 80% backup traffic destined to the Windows server across the teamed 10Gig NICs. Anyone know what the Properties of the NC522SFP should be in a scenario like this?

[Chief Rocker]

got the info that any server over 32GB would have an issue with the utility - dropped server down to 12GB, and nxflash_x86 still doesn't run.

[ABT]

HP NC510C losing connection
Hi Guys, im having the same issue over and over again with his card NC510c. ive dropped down from 32 RAM to 24 and still no help. Ive opened a case iwth HP on it and no idueas either, this card is a real piece of work. any input id appreciate it. my card drops after 2 hours.

[Chief Rocker]

OK...here is what I had to do:

1. drop the configuration to base - 8GB of RAM

2. Boot in safe-mode with networking support. If you boot without networking support, Windows can't see the card.

3. Run the firmware update utility.

This was for the NC375i, but assume the same process will work for the NC510C

[ABT]

Thanks, Ill try anything to fix this crap!
i put in a change control and going to do it now.. and let you know.
This is my Backup server, ive lost over 10 TB of backups because of this crap!
when you say FW update, which FW nd How?

[ABT]

per HP just now...

refer to KB945977

The problem may be the result of combining the NIC driver, and the TCP Chimney feature that comes with the Scalable Networking Pack, which is included in Windows 2003 SP2.

The TCP Chimney feature is a stateful offload that is enabled to be the default in Windows Server 2003 Scalable networking Pack. TCP Chimney offload enables TCP/IP processing to be offloaded to network adapters that can handle the TCP/IP processing in hardware.

With TCP Chimney feature, some NIC driver cause high NonPaged pool usage. The tag is MmCm.

Once there is less than 30MB of available non paged pool memory, HTTP.sys starts rejecting connections.

Contact the NIC manufacture to upgrade the NIC driver to the latest version. If the latest driver still does not resolve the issue, use the workaround below.



Workaround:

Disable the TCP Chimney feature on the server using the Netsh.exe tool. Specifically, to turn off TCP Chimney by using the Netsh.exe tool, we followed these steps:

1. Click Start, click Run, type cmd, and then click OK.

2. At the command prompt, type:

Netsh int ip set chimney DISABLED

3. Press the ENTER key.


For more information on the TCP Chimney feature and the Scalable Networking Pack,
see the following article:

912222 The Microsoft Windows Server 2003 Scalable Networking Pack release
http://support.microsoft.com/default.aspx?scid=kb;EN-US;912222 (http://support.microsoft.com/default.aspx?scid=kb;en-us;912222)


Query words: NetXtreme 2003 SP2 Scalable Networking Pack MmCm Tag TCP () Chimney () broadcom KB944011 944011

[Marcus Daubaras]

I know this is an old thread, but anyone getting the NetUserGetInfo:failed message should try to logon with a local account instead of a domain account. Worked for me.