NetBackup™ for Cloud Object Store Administrator's Guide
- Introduction
- Managing Cloud object store assets
- Planning NetBackup protection for Cloud object store assets
- Prerequisites for adding Cloud object store accounts
- Configuring buffer size for backups
- Permissions required for Amazon S3 cloud provider user
- Permissions required for Azure blob storage
- Permissions required for GCP
- Limitations and considerations
- Adding Cloud object store accounts
- Manage Cloud object store accounts
- Scan for malware
- Protecting Cloud object store assets
- About accelerator support
- About incremental backup
- About dynamic multi-streaming
- About policies for Cloud object store assets
- Planning for policies
- Prerequisites for Cloud object store policies
- Creating a backup policy
- Policy attributes
- Creating schedule attributes for policies
- Configuring the Start window
- Configuring the exclude dates
- Configuring the include dates
- Configuring the Cloud objects tab
- Adding conditions
- Adding tag conditions
- Examples of conditions and tag conditions
- Managing Cloud object store policies
- Recovering Cloud object store assets
- Troubleshooting
- Reduced acceleration during the first full backup, after upgrade to version 10.5
- After backup, some files in the shm folder and shared memory are not cleaned up.
- After an upgrade to NetBackup version 10.5, copying, activating, and deactivating policies may fail for older policies
- Backup fails with default number of streams with the error: Failed to start NetBackup COSP process.
- Backup fails or becomes partially successful on GCP storage for objects with content encoding as GZIP.
- Recovery for the original bucket recovery option starts, but the job fails with error 3601
- Recovery Job does not start
- Restore fails: "Error bpbrm (PID=3899) client restore EXIT STATUS 40: network connection broken"
- Access tier property not restored after overwriting the existing object in the original location
- Reduced accelerator optimization in Azure for OR query with multiple tags
- Backup failed and shows a certificate error with Amazon S3 bucket names containing dots (.)
- Azure backup jobs fail when space is provided in a tag query for either tag key name or value.
- The Cloud object store account has encountered an error
- The bucket is list empty during policy selection
- Creating a second account on Cloudian fails by selecting an existing region
- Restore failed with 2825 incomplete restore operation
- Bucket listing of a cloud provider fails when adding a bucket in the Cloud objects tab
- AIR import image restore fails on the target domain if the Cloud store account is not added to the target domain
- Backup for Azure Data Lake fails when a back-level media server is used with backup host or storage server version 10.3
- Backup fails partially in Azure Data Lake: "Error nbpem (pid=16018) backup of client
- Recovery for Azure Data Lake fails: "This operation is not permitted as the path is too deep"
- Empty directories are not backed up in Azure Data Lake
- Recovery error: "Invalid alternate directory location. You must specify a string with length less than 1025 valid characters"
- Recovery error: "Invalid parameter specified"
- Restore fails: "Cannot perform the COSP operation, skipping the object: [/testdata/FxtZMidEdTK]"
- Cloud store account creation fails with incorrect credentials
- Discovery failures due to improper permissions
- Restore failures due to object lock
About incremental backup
NetBackup supports incremental backups for Cloud object store workloads. You can use incremental backup without enabling accelerator.
For Cloud object store workload, some metadata properties do not alter the modification time for an object or blob. For example, the Tags in Azure blobs. Even if you change these metadata properties, the corresponding objects are not considered for the next incremental backup. This may appear as a loss of data during an incremental backup.
For Azure Data Lake and Azure Data Lake Government providers, when you update the ACLs of files or directories, the last modified time of the file or directory doesn't change. So, only by changing the ACLs, the files and directories do not qualify as incremental backups.
For a detailed list of metadata properties that do not alter modification time for an object or blob, refer to the respective cloud provider's documentation.
For incremental backups, if an object name has a path-style naming scheme, then for each path, an entry is added in NetBackup. If the object, which is represented by the end node of this path style naming, has not changed since the last backup (either full or last incremental, based on the incremental schedule used), then that object is not included in the next incremental backup. Because of this behavior, empty paths show up in the catalog and are rendered in the browse view of restore.