Restore failed with 2825 incomplete restore operation
Not all objects are restored from the backup image. Restore failed with 2825 incomplete restore operation.
Explanation:
This error can occur for multiple reasons. The most likely scenario for this error is when a cloud API initiated by NetBackup during a restore returns an error like an HTTP 400 status code (Bad Request). The reasons can vary with each cloud vendor. For example, GCP supports different Content-Language metadata as compared to AWS. In some cases, the error can also occur depending on the features enabled or disabled on a specific cloud account or bucket.
nbcosp log shows the following message:
{"level":"warn","error":"InvalidArgument: Invalid argument.\n\tstatus code: 400, request id: , host id: ","object key":"meta-user-defined/t2.rtf","time"…}
nbtar log shows the following message:
15:56:15.739 [22496.22496] <16> operation_to_cloud_by_type: ocsd reply with error, error_code: 400 15:56:15.739 [22496.22496] <16> CloudObjectStore::InitMultiPartUpload: operation_to_cloud_by_type() failed, status=3600 15:56:15.739 [22496.22496] <16> CloudObjectStore::ObjectOpen: InitMultiPart Upload call failed with status = 3600 15:56:15.739 [22496.22496] <16> cCloudApiRestoreHandler::writeOpen: ERR - ObjectOpen failed with error code [3600]
Workaround:
When the error is not fatal, the restore job is a partial success. Check the Activity Monitor to see the list of objects that cannot be restored. Try restoring to a different location (bucket/container or different account) to check if the problem is with the destination cloud account or bucket settings.
When the error is fatal, the restore job fails. Check the nbcosp logs to determine the object for which the restore has failed. Use granular object selection for the next restore, and skip the earlier failed object while selecting the objects.
Refer to your cloud provider documentation to check if you use a feature or any metadata that the cloud vendor does not support completely, or if it needs any more configuration. Fix the object with the right attributes in the Cloud object store and start a new backup job. Once this backup completes, the objects can be restored without this workaround.