HP NonStop RDF, Руководство по управлению

Страница: 171 / 478

through M and a new table
(tableB
) contains the keys N through Z. Suppose also that you
performed this operation manually on the backup system.
If you then initialize the RDF subsystem to a point in the MAT prior to the Stop-RDF-Updater
record associated with the partition boundary change and an updater encounters audit records
associated a key N through Z, the updater will report an error because it will try to apply the
audit records to
tableA
(which used to contain it, but now does not), and the audit records will
not be applied to the backup database. In this particular case the database is not corrupted, but
data corruption could occur for other NonStop SQL/MP or NonStop SQL/MX DDL SHARED
ACCESS operations.
If you did recently perform a NonStop SQL/MP or NonStop SQL/MXoperation with SHARED
ACCESS on the primary system and you want to initialize the RDF subsystem to a
synchdbtime
,
you must specify a
synchdbtime
such that the starting position in the MAT is after the
Stop-RDF-Updater record.
As a precaution, if RDFCOM encounters a Stop-RDF-Updater record during its backward search
of the MAT, it issues a warning message to that effect asking if you want to proceed with
initialization. If the extractor encounters such a record while operating in database synchronization
mode, it abends.
TMF Shutdown Records
TMF shutdown records in the MAT do not cause a problem, except that RDF shuts down and
you must then restart it.
CREATE/LOAD Issues (Step 4, Method 1)
Create an empty set of duplicate key-sequenced and relative files and tables that do not have
SYSKEYs and clustering keys on either the primary or backup system. The duplicate tables and
files must not be audited.
The LOAD command only works on tables and files that are nonaudited. If you create the empty
duplicate tables and files as audited entities, you must then use FUP ALTER commands to turn
off their audit attributes before you can load them.
You can create these files and tables on either the primary or backup system, but you should
only create them on the backup system if you have sufficient Expand bandwidth between the
two systems to handle both the audit sent by the extractor and the data sent by the load operations.
Using load commands, populate the empty tables and files. For NonStop SQL/MP tables, use
SQLCI LOAD commands with the SHARE option. For Enscribe files, use FUP LOAD commands
with the SHARE option.
For information about the SQLCI LOAD, MXCI LOAD, or FUP LOAD commands, refer to the
SQL/MP Reference Manual, the SQL/MX Reference Manual, or the File Utility Program (FUP) Reference
Manual, respectively. The following information is general in nature and is not intended as a
substitute for the information contained in those manuals.
General Considerations for Enscribe Files
• Key-sequenced files. To improve the performance of the load operations, specify the SORTED
option.
• Relative files. To ensure complete consistency with the source files, specify the NO
COMPACT option.
• Entry-sequenced.Currently create/load method is not supported for entry-sequenced files
and the only other way is to follow Method 2 discussed in the
“Overview” section.
• Unstructured Files. Unstructured files need to be synchronized offline.
Synchronizing Entire Databases Online 171