Hyper-V Cluster Shared Volumes (CSVs) are being left in a "Backup in progress, Redirected access" state.

Article:TECH205414  |  Created: 2013-04-23  |  Updated: 2013-09-30  |  Article URL http://www.symantec.com/docs/TECH205414
Article Type
Technical Solution

Product(s)

Issue




Hyper-V Cluster Shared Volumes (CSVs) are being left in a "Backup in progress, Redirected access" state.

This can also present itself as CSVs being left in "Redirected access" state.

Additionally, it can appear that Backup Exec Remote Agent process is hung on the Hyper-V node that owns the CSV, and Hyper-V snapshot jobs are hanging in the Backup Exec Activity Monitor.
 

 


Error



-Backup jobs could hang on the snapshot create/delete operation.
-Cluster Shared Volumes (CSVs) will be stuck in an Online (Backup in progress, Redirected access) state.
-CSVs will be stuck in an Online (Redirected Access) state.
-It is possible that a cluster node will create a STOP error and blue screen as part of a corrective action to process hang/crash.

 


Environment



System Center Virtual Machine Manager (SCVMM) 2012 with Dynamic Optimization enabled.
Backup Exec 2010 R3 or 2012.

 

 

 


Cause



If System Center Virtual Machine Manager (SCVMM) 2012 is installed, Dynamic Optimization may be configured.

Dynamic Optimization is a new feature in SCVMM 2012 that will dynamically migrate the virtual machines between cluster nodes for load balancing purposes.

More information about Dynamic Optimization can be found here:
http://technet.microsoft.com/en-us/library/gg675109.aspx

What was experienced in the Hyper-V environment was the following:

1) The guest VMs would be migrated to different Hyper-V hosts within the cluster in the middle of a backup job.
2) This would result in the snapshot for the guest VM being created on one cluster node (the one that originally hosted the guest VM), but the call to delete the snapshot would go to a different cluster node (the new host for the guest VM).
3) This caused the Backup Exec remote agent process on the original cluster node to wait for an indefinite amount of time for a snapshot delete job to run, and the Cluster Shared Volume (CSV) that stores the guest VM would remain in an "Online (Backup in progress, Redirected access)" state.  This Backup Exec remote agent process would appear to the user to be in a "hung" state, thus prompting the user to manually kill the process.
4) When the remote process was manually killed from Task Manager, the CSV would then move into an "Online (Redirected access)" state and the snapshot that was previously created would remain on the CSV.
5) While in this state, the CSV can change owners (cluster node) and snapshots from other guest VM backups could be created.
6) The next snapshot for a guest VM created would be created successfully, the backup run successfully, the snapshot would then be deleted, but the CSV would hang trying to clear the "Redirected access" state.

This API hang to clear the CSV from "Redirected access" state may cause the cluster node to generate a STOP error and blue screen as part of a corrective action by Microsoft.
http://msdn.microsoft.com/en-us/library/windows/desktop/aa369104(v=vs.85).aspx


 
 


Solution



From within the System Center Virtual Machine Manager, disable Dynamic Optimization.

 





Article URL http://www.symantec.com/docs/TECH205414


Terms of use for this information are found in Legal Notices