We are frequently encountering a problem where data chunks are missing, and this is impacting our ability to restore multiple VMs. Initially, we suspected an issue with the cloud storage provider and requested the support team to investigate. However, upon further analysis, we discovered that a delete command was issued by Commvault, which led to the deletion of the data from cloud storage.
As a result, multiple storage policies have been affected, and we are currently unable to restore the data for several virtual machines.
Analysis:
According to DataCore logs:
- The SFILE delete request was received at 08:08:49
- The Compact File delete request was received at 08:07:55
Commvault is expected to send a delete request for the Compact File only after a successful remote copy of the Compact File to recreate the SFILE. If this process is followed correctly, the SFILE should have already been deleted before the Compact File is removed.
However, this sequence appears to be reversed, and DataCore has not shown two delete requests for the SFILE. This raises concerns about the deletion order and whether Commvault is handling it correctly. The issue remains unresolved.
Questions:
We are seeking clarification on how Copy-by-Range and Remote Copy requests are processed.
Based on documentation and feedback from trainers and support engineers, these steps are critical for Space Reclamation. Specifically:
- During these copy processes, is the data temporarily downloaded and uploaded to the cloud?
- Is this the reason a Media Agent is required on the cloud?
While this question may not be directly related to the current issue, we are trying to better understand the underlying process.
