RDF System Management Manual for J-series and H-series RVUs (RDF 1.10)

also create them on the backup system. You then must purge the files on your backup system before
copying the loaded files from the primary system.
General Considerations for NonStop SQL Tables
• Key-sequenced tables without SYSKEY. To improve the performance of the load operations
significantly, specify the SORTED option.
• Tables with SYSKEY or Clustering Keys. Because the NonStop SQL load operations generate
new SYSKEY and clustering key values, do not use the create/load method to get a copy of
such table. Instead, use the method explained in Step 4, Method 2 under “Synchronizing
Entire Databases Online” (page 159) .
• Relative tables. To ensure complete consistency with the source files, specify the NO COMPACT
option.
• Entry-sequenced tables. Do not use the create/load method to obtain a copy of entry-sequenced
files. Instead, use the method explained in Step 4, Method 2 under “Synchronizing Entire
Databases Online” (page 159).
• Partitioned tables. You can initiate the load operation with a single command by executing
the LOAD command against the primary partition.
• Index tables. With regard to index tables, there are several considerations for relative and
key-sequenced tables. Regardless of base table type, you cannot load an index table by itself.
Index tables can only be loaded when the associated relative or key-sequenced table is loaded.
If you want the associated index tables loaded when you load a relative or key-sequenced
table, you must create empty index tables first, before issuing the LOAD command. When
you load the base table, the index tables are loaded automatically.
Alternatively, you can load your base tables without index tables. Then you can create and
populate your index tables with the NonStop SQL product of choice before you start the RDF
updaters.
If you fail to create the index tables before issuing the START UPDATE command, the affected
updater reports a file-system error 11 (File not found) when trying to apply an update to an
index table, and it continues to retry the update. In this situation, the updater does not make
forward progress until you create the index on your backup system.
Enscribe Queue File Issues
For ENSCRIBE queue files, a different method of obtaining the fuzzy copy is required. You must
use the FUP COPY command with the SHARE option specified, and with “FIRST 1” specified. For
example, the following command copies the contents of file QUEUE1 to QUEUE2.
FUP COPY QUEUE1, QUEUE2, FIRST 1, SHARE
To ensure that your target file, QUEUE2 in the above example, has the proper content, copy the
content of the target file to the screen by using the following command:
FUP COPY QUEUE2,, H
If the file is empty and contains zero records, you must reissue your original command again, and
recheck the contents of the target file.
FUP COPY QUEUE1, QUEUE2, FIRST 1, SHARE
FUP COPY QUEUE2,, H
The target file, QUEUE2 in this example, is not ready for synchronization until it has at least one
record in it. Therefore, you might need to repeat the above operation until a record appears.
You could also copy the empty file to the backup system, insert a record into the file on the backup
system, and delete the inserted record:
1. Start a transaction, do a WRITE to the empty queue file, and commit the transaction.
Synchronizing Entire Databases Online 163