Subscribe

Cloud reclamation issues with SteelStore AMI 2030

We have a SteelStore AMI 2030 on AWS. We are facing issues with cloud storage reclamation.

The reclamation cycle shows completed. However, the utilization S3 bucket doesn't come down.

Any help on this would be appreciated.

Re: Cloud reclamation issues with SteelStore AMI 2030

Hi,

It is possible that the reclamation process runs but doesn't have any space to reclaim because the data is all references and not actual data. What actions are you taking to remove data on the front end (backup side) of the SteelStore AMI, which would trigger the cloud space reclamation? Reclamation should work fine and most users typically see a plateau when the backup policy reaches steady state (with small bumps and valleys as backups are taken and old ones expired off SteelStore).

 

Regards,

Christopher

Re: Cloud reclamation issues with SteelStore AMI 2030

We are using Symantec NetBackup as the backup solution. NBU on its own expires the data once the retention period is over.

We suspect that there is some data on the S3 bucket which has been already expired NBU, however, that has not been deleted from the S3 bucket. One of the reasons for this suspicion is that reclamation was not working on this AMI/appliance and we had manually triggered it.

Re: Cloud reclamation issues with SteelStore AMI 2030

When NBU (or any backup application) removes old backups, those are recieved as deletes for files on the SteelStore CIFS/NFS share. The corresponding deduplicated data or references are deleted by SteelStore, and as appropriate the cloud storage slabs which hold the deduplicated data are reorganized or deleted according to what is removed and no longer relevent. It seems that in this case, the deletion of the data must be mostly references and not data.

 

If you feel this is not the expected behavior, then we may need to perform further investigation via a support ticket to NetApp support.

Re: Cloud reclamation issues with SteelStore AMI 2030

We already have a support ticket open and engineering is investigating this particular issue