Failed backups are not identified as corrupted

Backups that never complete (e.g., sql server hangs and never replies) will leave behind partial backup files that are not identified as corrupted.

I would propose that backup files are created in the "corrupted" folder, and moved out to the proper backup directory only after they have been verified as complete. A corrupted until proven valid mindset if you will...

For some more background on how we've discovered this issue, we've uncovered an issue with the VM's we are running in that there is an abnormally high percentage of packets being dropped. This is causing random queries to hang when using JIRA. The backup system in this case causes a very high number of queries, making it almost guaranteed that the query system will hang during the backup process. I have used remote debugging in Eclipse to confirm that the particular backup thread ends up hung waiting for data from the SQL server (readSocket0).