Hi,
I am running in to periodic issues with Windows Server Backup backing up deduplicated data to a UNC path. Both Servers are domain joined and running 2012 DataCenter edition.
The backup process does not begin logging Event ID 153 until it reaches the deduplicated volume. It successfully backed up its own system state and C drive volume (~21GB), but upon reaching the deduped volume, began throwing IO errors. The IO errors are on the mounted UNC volume to which the server backing up copies it backed up data.
Once the errors began spewing out, the backup will eventually stall. The most reliable manner for me to recover the server running the backup is to block the UNC traffic at the server hosting the backup data volume. This action routinely and repeatedly allows the backup process to terminate. Normally, I cannot successfully terminate the backup process by cancelling.
What suggestions might any one have for me to more closely analyze the filesystem operations on the deduped volume conducted by wbengine? I ran procmon before the wbengine process stalled, and did not see any precise process activity that corresponded to the Event Viewer entries.
After the wbengine process stalls, procmon does not load. However, if I enforce my cut-off with the Windows Firewall on the backup data server, procmon will load following the successful termination of wbengine.
Any help is appreciated. I have a peer with a similar issue happening, although that issue has a directly mounted iSCSI backup volume yet the source volume for the backup is also deduped.
Michael