increasing amount of BSODs of elxcna.sys

[SwisspostIT]

hello,

starting a few weeks ago we have an increasing number of blades which are affected by bluescreens.

The Tool Bluescreen View allways points to the Emulex CNA FCoE Driver "elxcna.sys".

We've opened today a case at HP Support but it would interest me if other people are observing this behaviour...

Here the Details of our equipment:

we have about 40 C7000 Enclosures filled with BL465c G7 and BL460c G7 servers. every enclosure has two HP 10GbE Pass-Thru Modules as interconnect connected to Cisco Nexus 2000 Top of the Rack Switches.

We are using FCoE for SAN connection and use the Onboard Emulex CNAs (NC551i and NC553i).

Last week we updatet about 20 Servers with the new CNA Firmware 4.0.360.15 and also the Drivers (NIC: 4.0.360.8 / FCoE: 7.2.50.007) but on friday we had 4 servers which had a BSOD again even with new Firmware and drivers...

The strange thing is that the BSODs even occur on bladeserver which doesn't have any SAN Discs zoned (only the FCoE NIC personality feature activated on the adapter through BIOS)...

Server OS is Windows Server 2008 R2

Does anyone have same issues on c-Class Blades?

Best Regards,

Ville

[DGauld]

Just wondering if this issue was ever resolved?

We just purchased a C7000 encloser and a BL460c G7 and after the first week of setting things up came across this same issue. Kind of at a loss of what is causing this or why.

Any advise you might be able to provide would be greatly appreciated.

Thanks

[JimP2]

I am having that same issue with a C7000 and a BL460c G7.  Can't seem to get around it yet.  LEt me know if anyone has a fix.  Cleared NVRAM, changed Mem modules.

[JimP2]

Folks,

I worked with HP today.  My server is a simple Win2K8-R2 server hosting Windows 2008 storage server. Basically a simple NFS share.   I tried changing the drivers, clearing NVRAM, different memory dimms. 

The one thing HP pointed out to me was the HP Networking Manager, IE: the SW that handles network port teaming on the local machine. They requested we uninstall this software...  We were not using the teaming so uninstalling it was not an issue.  Once uninstalled this server has not Blue Screened as of yet with the ELXCNA,SYS driver/IRQ Not LESS OR EQUAL error.   HP is also sending out a replacement system board for this server.  The last option they gave me was to reinstall the OS, which i know my MGMT will not want to hear.

Anywho, i hope this helps some of you with your similar issues.  I will update next week with a status of this next week.

-jimP

[JimP2]

My OS (W2K8-R2) is updated with the latest patches from HP and MS. It is also the 64bit version.

[SwisspostIT]

Hi everyone,

at our site the problem disappeared without knowing why...

We had for about 2-3 months these BSODs but someday they disappeared (we didn't really change anything as much I know).

At the moment we are running CNA Firmware Version 4.1.402.20 and elxcna.sys version 7.2.40.008 on our Windows Server 2008 R2 Blades (BL465c G7 and BL460c G7) and as far as I know we didn't have any more BSODs.

BUT: we have some ESXi clusters of BL685c G7 servers where some times a host suddenly makes an unexpected reboot... We're working on this with HP since about 7+ months know and didn't find any solution until know. But in this environment we also use Flex-10 Switches and the unexpected reboots point rather to CPU than to CNA causing this.

Best regards,

Ville

[dupre299]

We are having the same problems too. We have 12 new DL380-Gen8 servers each with a CN1100E and 554FLR FlexLOM card in it. They are Emulex BE3 based boards. All latest firmeware on everything dated July 2, 2013. The Emulex elxcna.sys FCoE Driver version is 2.74.9.1, the network driver (not that I believe it is relevant) is 4.2.390.6. The firmware revision is 4.2.401.2215 (dated June 2013 from HP's site).

We have a LBFO team in Windows 2012 connected using an LACP Switch Dependent connection to two Nexus 5548UP - the port channel is a VPC as well to allow cross-Nexus port channeling. 1 Link from each adapter is connected in the port channel, the other two ports (1 from each adapter) are disabled.

What seems to trigger the blue screen is when we make a change on the port config on the Nexus - but it is not always able to be replicated. For instance, we changed a native VLAN one day - it blue screened. We changed the allowed VLANs one day deliberately to remove the FCoE VLAN to force a blue screen ... and it did. Then did the same exact sequence of events on the Nexus, no blue screen. The blue screens are always pointing to the elxcna.sys and are a Stop 0x00000139 with the next number being 0x0000000000000003. This is not an OS issue, and MS will not allow this to be a case they work on because the blue screen points to the elxcna.sys driver. Unfortunately, Emulex will not work on this either because these are HP branded Emulex cards...

HP are you listening?

Case number 4645538594.

So we have a half million dollars worth brand new boat anchors right now...

Very frustrating!

[dupre299]

UPDATE: This is a known issue with the Emulex drivers currently on the HP website. FCoE driver 2.74.09.1 has this bug according to Emulex. This issue has been escalated to L3 at HP to correct. According to Emulex, 2.74.14.1 has corrected this bug in their generic driver set. HP needs to certify this driver and is currently testing 2.74.16 in QA for release in August. I will post as I get more updates. Good news is that it seems to be a known issue and HP just needs to release the already produced driver set from Emulex that corrects the issue.

[dupre299]

Update: This has been corrected in the September 10 release by HP of FW and drivers for Emulex CNA (CN1100, Flex 554, etc).