SQL/MP Installation and Management Guide

Performing Recovery Operations
HP NonStop SQL/MP Installation and Management Guide—523353-004
11-28
Recovery Example
must alter the dumps of the purged objects to reset the INVALID and RELEASED flags
before attempting recovery.
You can recover SQL objects to a different location even when the source objects have
not been purged. You can create the target objects to match the source, and then
perform recovery to obtain a copy of your source objects. Transactions can be active
against the source objects at the time of recovery.
Other Recovery Methods
Other methods of recovering a dropped SQL table are possible, but they are riskier
than the method just described. For example, you could use the same TMFCOM
RECOVER FILES command to recover the SQL catalog as well as the table and its
dependent objects. If successful, this method would eliminate the need to synchronize
the timestamps in the catalog with those in the file labels, but note:
•
You must recover the catalog precisely up to the time of the SQL table drop.
Recovering to a later time causes you once again to lose the entries for the table.
•
If you recover the catalog to the time of the table drop and if any updates were
done to the catalog after the drop, the resulting catalog would miss those updates.
Therefore, you should definitely avoid this method if it is possible that a catalog
update occurred after the table drop. Such a step could create even more
inconsistencies between the SQL catalog and the SQL objects. Consequently, you
Caution. Use the TOFIRSTPURGE, TIME, or TOMATPOSITION option in the RECOVER
FILES command to avoid replaying the purge operation on a target object. If you do not do this
and the file-recovery process encounters a purge record for an object being recovered to a
different location, the process terminates recovery of that object with these EMS messages:
NonStop TMF on \PLUTO *0437* RECOVER FILES [60]
FileRecovery Process #1: Encountered a purge record
for audited file $DATA16.PERSNL.EMPLIST while performing
FLABMOD REDO operation; Audit Trail Index #2, SNO #1,
RBA #22707360.
NonStop TMF on \PLUTO *0358* RECOVER FILES [60]
FileRecovery Process #1: Recovery on $DATA16.PERSNL.EMPLIST
has terminated.
Caution. If you use the MAP NAMES option of the RECOVER FILES command to recover
files to a new location, you must immediately make new online dumps of the target data files
recovered. Without these new dumps, you will not have file-recovery protection for those files,
and subsequent file recovery operations can fail. In particular, if you later try to use old online
dumps of the target files to recover the target files to a point beyond the time that the last
RECOVER FILES command was issued, the file recovery process fails during the redo phase
and transmits EMS message 175:
Encountered a File Hiatus record for audit file filename at
audit trail Index #index, SNO #sno, RBA #rba.