Diagnosing Windows' IO_DEVICE_ERROR 1117 (0x045d)

alx9r · Mon Aug 25, 2014 5:46 pm

I am using Starwind iSCSI SAN and NAS to provide an iSCSI connection to a direct-attached tape library. Here are the particulars of the setup:

Code: Select all

Tape Library: Dell TL2000
    ^
    |  SAS cable
    v
Dell SAS 6Gbps Host Bus Adapter
    ^
    |  PCIe
    v
Dell PowerEdge R720xd Running Windows Server 2012R2 as a hypervisor
Starwind iSCSI SAN v6.0.0 is running here and exports the two TL2000 scsi device as iSCSI targets
    ^
    |  host virtual network adapter
    v
Hyper-V "internal" virtual switch
VMQ and SR-IOV enabled
Phyical Adapter is Intel X520-DA2
    ^
    |  VM virtual network adapter
    v
Windows Server2012R2 VM hosted on above hypervisor
Tape Drive Driver: Microsoft's 2006-06-21 LTO Tape Drive v6.3.9600.16384
Running System Center Data Protection Manager (DPM) 2012R2

The above arrangement works very well with this small exception: About 30 minutes after the start of a tape backup job DPM reports error 0x8007045d which (as best I can tell) is caused by Windows reporting IO_DEVICE_ERROR 1117 (0x045d). The problem is that DPM interprets this as "end of tape" and moves on to the next tape when the first tape is nowhere near full.

I am new to troubleshooting SAS, iSCSI, and tape drives. Can you provide some suggestions to narrow down the cause of this error?

Thanks for your help,

Alex

Mon Aug 25, 2014 6:14 pm

Did you adjust transfer size like it's recommended? Can you provide complete StarWind log for a failed operation?

alx9r · Mon Aug 25, 2014 7:04 pm

Hi Anton,

Thanks for the quick response. This is the first I have heard of adjusting transfer size. Although I just read this and it makes sense to me: http://www.starwindsoftware.com/forums/ ... 089&start= Are those the most up-to-date recommendations on transfer size?

Here are the starwind logs for the problem event:

Code: Select all

8/22 22:02:19.678 12a0 Srv: Remove Complete '\\?\SCSI#Disk&Ven_Msft&Prod_Virtual_Disk#2&1f4adffe&0&000023#{53f56307-b6bf-11d0-94f2-00a0c91efb8b}'
8/24 3:00:50.888 1098 Tgt: iqn.2008-08.com.starwindsoftware:10.202.3.5-ace-ptl01-drv01: sess [6] LUN 0 * UNIT ATTENTION (0x28, 0x0)
8/24 3:03:52.138 1098 Tgt: iqn.2008-08.com.starwindsoftware:10.202.3.5-ace-ptl01-drv01: sess [6] LUN 0 * UNIT ATTENTION (0x28, 0x0)
8/24 3:19:41.944 1098 Tgt: iqn.2008-08.com.starwindsoftware:10.202.3.5-ace-ptl01-drv01: sess [6] LUN 0 * UNIT ATTENTION (0x28, 0x0)
8/24 3:32:29.379 1098 Tgt: iqn.2008-08.com.starwindsoftware:10.202.3.5-ace-ptl01-drv01: sess [6] LUN 0 * UNIT ATTENTION (0x28, 0x0)
8/25 9:36:01.876 ca0 conf: Accepted control connection from 10.202.3.57:2366.

The 0x045d errors occurred at 3:27:47 and 3:32:33 on 08/24.

Mon Aug 25, 2014 9:45 pm

Yes, these are correct ones. 64-128KB should work. Sorry we cannot do it on a client side as we don't run there...

Please don't post logs in body of the messages as they are getting truncated. Zip and send to support@starwindsoftware.com or attach to post. Thanks!

alx9r · Tue Aug 26, 2014 5:14 pm

I have change MaxTansferLength to 64kB and the errors have disappeared. Is there a manufacturer's specification or a tool that I can use to determine the highest transfer length that doesn't cause problems?

Mon Sep 01, 2014 9:36 am

I think you`ll find the information that you need here:
http://www.starwindsoftware.com/forums/ ... f=5&t=1089