Protection Manager can now perform an on-demand Backup using the DFM CLI that contains a subset of the members of a full backup. This ability now makes it possible for customers to backup specific primary dataset members to form a new backup version with just the members specified. The need for this ability was to allow a customer who has noticed that a backup was missing 1 or more members due to some equipment issues to do a new on demand backup so that the prior missing members will have been backed up.
This new ability is only available from the DFM command line interface and there is no GUI support for this partial on demand backup.The CLI itself does not check the member IDs specified so the job log is the only place that missing members or incorrectly specified member IDs can be detected.
The CLI command dfpm backup start allows the -o parameter to be used to supply comma separated primary node member IDs. This list will be used as the backup-members list that is issued as part of the backup start CLI. The CLI code itself does not check that the member IDs specified are part of the dataset. Therefore, no direct feedback comes to a user of this functionality. The user may check the member IDs using dfpm dataset list -m <myDatasetName> in order ensure correct member IDs are used.
The job that does the backup or mirror updates will be passed the list of root node members to be backed up directly as specified by the user. Any issues that have to do with incorrect member IDs will only be reported in the job logs and the text log files that are normally part of DFM debug information. Should the user enter a list of members where none of the members are in the dataset the job will run and contain one warning for each bad ID. A job where no members were found that were part of the dataset will end with a job status of failure.