Summary
RMAN looks for the datafiles to backup and identifies that the datafile doesn't exist any more ( dropped ) that cause the backup to fail.
The question is,
Does any workaround/patch exist so rman does not fail and exit, but to continue the backup with error message,
or warning that the Tablespace / Datafile was dropped after backup has started and so not available.
Error:
RMAN-03002: failure of backup plus archivelog command at 02/05/2013 12:51:00
RMAN-03009: failure of backup command on ch01 channel at 02/05/2013 12:51:00
ORA-01157: cannot identify/lock data file 5 - see DBWR trace file
ORA-01110: data file 5: 'No file with this number, file does not exist'
or:
RMAN-03002: failure of backup plus archivelog command at 02/06/2013 15:55:27
RMAN-06056: could not access datafile 5 <<-- This tablespace/datafile was dropped after rman backup was already started
Solution
Rerun the Backup !
Current the only workaround is to move the schedule of the backup so it falls outside the maintenance to 'drop' Tablespaces/Datafiles.
An Enhancment request if filed already in
<Bug 10405386>: MAINTENANCE TASK LIKE DROP TABLESPACE/DATAFILE DURING RMAN BACKUP
to change the behavior and let backup continue even though the dropped datafile no longer available.
Attachments :
Cause
This is expected behavior and it is not supported to drop/create tablespaces while performing a rman backup.
When rman starts a backup, it will take a snapshot of the controlfile (for read consistency) to get the list of datafiles to backup.
If during the backup, some of these files are dropped, rman will error out as it still expects these datafile(s) to be present.
The same will happen for newly created tablespaces (but will have no impact as such), they will not be part of the backup as they were unknown at the time the backup started.