ONBAR

 ONBARS FUNCTIONAL SPECIFICATIONS

OnBar Operations

Onbar has 5 types of backups: 

1. Whole System Backups - (onbar -b -w) A whole system backup is indicated with the '-w' flag in the onbar command. When this is set onbar will backup each dbspace serially and such that there will not be any need for logical recovery if a restore is required. A whole system backup will back up every dbspace in the instance.
2. Dbspace Backups - (onbar -b -L 0 dbsp1 dbsp2 ...) A dbspace backup is when onbar will backup a single dbspace or a group of dbspaces as specified by the argument list passed the the onBar command. If you are backing up multiple dbspaces, then the backups are done in parallel. The amount of backups that can be done at 1 time is restricted to the value of BAR_MAX_BACKUP (see onBar Configuration Parameters). You may also put the list of dbspaces you want to be backed up in the a file and has that file referenced by onBar (see the usage below). If a restore is done utilizing these backups then logical recovery is required to sync up these dbspaces with the other dbspaces.
3. Logical Log Backups - (onbar -l) onbar will request all logical logs from the engine to be backed up and pass them to the storage manager.
4. Continous Logical Log Backups - (onbar -C -l ) onbar will start continous logging just like ontape -c
5. Fake online Backup - (onbar -b -F) onbar will tell the engine to do a 'fake' backup. This is like archiving to /dev/null. This added feature is useful if you want to change the logging mode af a database and do not have the time to perform and archive. NOTE. This does not take the place a regular archive because nothing is sent to the storage manager so nothing can be restored.

Onbar has 6 types of restore options 

1. Onbar Cold Restore - (onbar -r) When onBar performs a cold restore it first attempts to do a cold log backup (salvage of any logical log that have not yet been backed up. Then onbar will scan the ixbar file to get a list of all objects it needs to restored (ie all level 0, 1, 2 backups of all dbspaces as well as all logical log that will need to be rolled forward). After that onbar simply get one of those objects from the storage manager and passes it to the online engine. once the physical restore is complete, onbar Will automatically start logical recovery and roll forward all logical logs it has.
2. Onbar Warm Restore - (Onbar -r dbsp2 dbsp3 ...) When onbar performs a warm restore it will first have online back up all the logical so that it can roll forward through them after the physical restore. Next onBar will scan through the sysutils database to get a list of the objects it needs to get from the storage menager. once that is done, onBar will simply get the objects from the storage manager and send then to the engine to be restored. Finally onbar starts retrieving the logical log backups from the Storage Manager and send them to online to be applied as logical recovery.
3. Mixed restore - Obar can allow the user to perform a mixed restore. This is basically a combinations of both cold and warm restores. The user would first perform a cold restore of the cirtical dbspaces and then a warm restore of the non-critical dbspaces.
4. Physical restore - (onbar -r -p) As mentioned above onbar will by default start logical recovery after physical recovery has finished. If the user does not want this to happen, he/she can use the '-p' option and onbar will only perform physical recovery. once this is done, online will hang in fast recovery until onmode -m is issued.
5. Logical recovery - (onbar -r -l) If the user does a physical restore and then wants to perform a logical restore after that, then the user can use the '-l' option to rollforward the logs after the physical restore completes. This can't be done if modifications were done to the database between them time the physical restore completed and the logical restore startted.
6. Point in Time Recovery - (onbar -r -t {date and time} OR onbar -r -n {log number}) This will allow the user to restore the ENTIRE instance to a specific point in time. That time can be specified by either telling onbar what logical log to rollforward through or by telling onbar the time it is to stop rolling forward. NOTE: The point in time recovery feature is only to be used when performing a cold restore of the entire instance and will NOT work if you want to perform a warm restore of a set of dbspaces.