About restoring data from Amazon Glacier
The NetBackup image is stored as set of objects with specified storage class, in this case, GLACIER or GLACIER_DEEP_ARCHIVE storage class. Restore from Amazon Glacier happens in two phases:
The objects are first retrieved at an internal staging location that Amazon maintains.
From there, the data is restored at the destination location.
NetBackup supports the following Amazon retrieval types:
Bulk retrieval, which completes within 5-12 hours.
Standard retrieval, which completes within 3 - 5 hours.
Expedited retrieval, which completes within 1-5 minutes.
For more about Amazon S3 storage classes, review Amazon S3 Storage Classes.
Note:
If you specify Expedited retrieval, Amazon can sometimes fail the request because of a lack of resources. If this failure happens, you must use Standard retrieval or Bulk retrieval. In this case, the restore job fails (NetBackup status 5: restore failed completely).
The activity monitor displays this message from bpbrm: Image warming failed 503. The following error is in the ocsd_storage log on the MSDP server when MSDP direct cloud tiering is used:GlacierExpeditedRetrievalNotAvailable: Glacier expedited retrievals are currently not available, please try again later status code: 503
When you perform a restore, the entire image fragment is restored while only the selected objects are downloaded.
Note:
If you use Glacier with MSDP direct cloud tiering, you can create GLACIER_RETRIEVAL touch file on master server in /usr/openv/netbackup/bin directory with one of three strings in it: bulk, standard, or expedited. You can create this touch file if you do not want to use the Bulk retrieval option.
If you use Glacier then you can use bulk, standard, or expedited. If you use DEEP_ARCHIVE you can use bulk or standard. If no string is defined, NetBackup's default is bulk if the touch file does not exist.
If you use Glacier with standard, non-deduplication cloud storage servers, only Amazon Standard retrieval is supported.
For more about restoring using Amazon S3, review Restoring Archived Objects.
Note:
This section does not apply to MSDP direct cloud tiering. The section only applies to standard, non-deduplication cloud storage servers.
If the files and folders you want to restore belong to multiple image fragments consider the following:
One image fragment is retrieved at a time. Only after the selected files and folders part of the first image fragment are downloaded, the next image fragment is retrieved.
The restore time must be considered depending on the number of image fragments. For example, if the files you want to restore are part of two fragments, an additional 6 - 10 hours are added to the complete restore time.
Note:
If you cancel a job after the restore retrieval is initiated, cost is incurred for all the objects that are retrieved on the staging location till the point of cancellation.