"failed on physical path" after creating RDM lun via Snapdrive

[daniel_kaiser]

All,

I'm trying to troubleshoot an issue that i'm noticing after i create an RDM lun via Snapdrive in a Windows 2003 VM host.  I created a LUN via Snapdrive on my VM guest machine without any issues.  I can see the LUN just fine via Snapdrive in Windows.  The problem is when i reboot the VM machine, i see the below messages only when the machine powers on and the OS loading.  Can someone explain why these errors are showing up, what it means, and how to resolve it.  Thanks.

ESX 4.0 update 1

VM guest machine:  windows 2003 r2 x32

SnapDrive for Windows 6.2

Error messages from the vmkernel log file.

Oct 14 15:20:35 wdcesx3 vmkernel: 0:06:53:45.778 cpu11:4502)VSCSI: 3472: handle 8234(vscsi0:1):Creating Virtual Device for world 4503 (FSS handle 209502576)
Oct 14 15:22:45 wdcesx3 vmkernel: 0:06:55:55.910 cpu2:4446)NMP: nmp_CompleteCommandForPath: Command 0x3c (0x41000c225c80) to NMP device "naa.60a98000572d4f6e4a5a5a645850354d" failed on physical path "xsigo_vHBA_2:C0:T1:L33" H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x20 0x0.
Oct 14 15:22:45 wdcesx3 vmkernel: 0:06:55:55.910 cpu2:4446)ScsiDeviceIO: 747: Command 0x3c to device "naa.60a98000572d4f6e4a5a5a645850354d" failed H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x20 0x0.
Oct 14 15:22:45 wdcesx3 vmkernel: 0:06:55:55.910 cpu2:4446)NMP: nmp_CompleteCommandForPath: Command 0x3c (0x41000c1337c0) to NMP device "naa.60a98000572d4f6e4a5a5a645850354d" failed on physical path "xsigo_vHBA_2:C0:T1:L33" H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x20 0x0.
Oct 14 15:22:45 wdcesx3 vmkernel: 0:06:55:55.910 cpu2:4446)ScsiDeviceIO: 747: Command 0x3c to device "naa.60a98000572d4f6e4a5a5a645850354d" failed H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x20 0x0.

[watan]

Did some searching and found the following on a internal Burt which had some good info. 
It looks like its a command that OnTap doesn't support and the error is produced.

0x3C Read Buffer      0x05/0x20/0x00 OnTap does not have any buffer for them to read.
We have no plans to have a buffer to ever support this command.

[daniel_kaiser]

Allan,

Thanks for replying to both this post and the one in the forum.  I posted the one in the forum yesterday, but somehow i thought i posted it in the communites, hence the new post in the communites.

Anyways, I really don't understand what you mean by the command isn't supported via Ontap.  What command are you referring to?

Also, i just completed building a new test VM running Windows 2008 R2 x64 and installed SnapDrive for Windows 6.2 x64 and didn't see any failed paths issues after booting up the VM that has RDM LUNS.  Both the luns are on the same filer.  Any ideas why i would see the errors on the 2003 host and not the 2008 host?

[watan]

Sure not a problem.

The reference I noted had the exact error and mentioned that the VMkernel was sending a command "nmp_CompleteCommandForPath: Command 0x3c" which translates to a "read buffer" which isn't supported in OnTap.

Not sure why you're getting it in w2k3 and not w2k8 but there has been recent activity that had issues with w2k3 & ESX 4.1

Perhaps a difference in drivers used?

Again not sure if its related but worth a read if you're on 4.1 w/ w2k3. -> http://now.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=447078

Also, I'd recommend opening a NGS case if its causing problems in your environment.  They'll have direct access to escalations if needed.