Notice: This website is an unofficial Microsoft Knowledge Base (hereinafter KB) archive and is intended to provide a reliable access to deleted content from Microsoft KB. All KB articles are owned by Microsoft Corporation. Read full disclaimer for more details.

When the Volume Shadow Copy Service performs an import procedure or a delete procedure on a Windows Server 2003-based computer, a SCSI device may not be correctly added to the PNP device tree


View products that this article applies to.

Symptoms

On a Microsoft Windows Server 2003-based computer, the Volume Shadow Copy Service (VSS) performs an import procedure or a delete procedure. However, a SCSI device may not be correctly added to the Plug and Play (PnP) device tree or removed from the PnP device tree. In this case, the following symptoms also occur:
  • One or more disk devices displays the 0xa (CM_PROB_FAILED_START) problem code in Device Manager.
  • The SETUPAPI log reports that a disk device did not start. The CM_PROB_FAILED_START problem code is logged.
  • VSS does not import or delete shadow copies during later backup attempts.

↑ Back to the top


Cause

This problem may occur if the following conditions are true:
  • Two calls are made to the CVssHardwareProviderWrapper: DeleteSnapshotsInternal() API in a 30-second interval.
  • A hardware provider removes a logical unit number (LUN) during the snapshot deletion.
  • The device that is being removed is attached to a host bus adapter that uses the SCSI Port miniport driver model.

↑ Back to the top


Resolution

The Storport storage driver model was designed by Microsoft to overcome the architectural limitations of the SCSI Port miniport architecture. We recommend upgrading to Storport-compliant host bus adapter drivers to resolve this problem.

↑ Back to the top


More information

When a shadow copy is imported or deleted, VSS determines whether the procedure caused a LUN to be masked or unmasked. If a LUN is masked or unmasked, VSS requests a rescan of all SCSI adapters on the computer. This rescan notifies the PnP Manager of the device changes.

The SCSI Port miniport architecture is designed to prevent multiple user mode rescans if they occur within a 30-second interval. Therefore, the SCSI Port miniport driver ignores any later rescan attempts that are initiated from user mode within the 30-second minimum rescan interval. PnP relies on the SCSI bus rescan to notify the system of SCSI device changes. Therefore, detached devices may continue to exist in the PnP device tree. You can see this behavior in Device Manager.

Note If multipath software is used, additional symptoms may occur. The PnP device tree may contain stale device objects for path devices that no longer point to a physical device. A wide variety of errors may occur if the LUN is re-introduced to the system after the multipath software has marked the paths as not valid.

↑ Back to the top


Keywords: KB935460, kbprb, kbtshoot

↑ Back to the top

Article Info
Article ID : 935460
Revision : 4
Created on : 10/11/2007
Published on : 10/11/2007
Exists online : False
Views : 308