Release-specific Upgrade Issues
Note the following upgrade issues that are specific to each release.
Ingres 11.0
• Binaries are not compatible with previous Ingres versions due to the introduction of the IANA time zones in Actian X 11.0. When initializing the runtime environment, applications may see the error “time zone files are corrupted.” You must recompile your applications.
• After upgrading to Ingres 11.0 with upgradedb, X100 operations are enabled in all user databases. The alterdb command flags -enable_x100 and -disable_x100 will enable and disable access to X100 tables and X100 operations for the database. You can permanently prohibit the use of X100 operations in a database using the -no_x100 flag on the createdb command. This prevents the X100 Server from starting, and the creation of X100 tables.
• If you are upgrading from an Ingres release to Ingres 11, have enabled X100 operations (this can be checked with infodb), and have created custom scripts to back up a database, the scripts must include the files and directories in the ‘vectorwise’ location. Failure to update your scripts to include all relevant directories may cause problems with checkpoints and recoveries. For more information on the vectorwise location, see Database File Locations in the Database Administrator Guide.
• A database upgraded from an earlier release of Ingres may contain tables that can cause E_DM9665_PAGE_OUT_OF_DATE and E_DM9666_PAGE_LSN_MISMATCH errors if updates to the table need to be aborted. Modifying the tables will prevent these errors from occurring.
To identify the tables that may be affected by this problem, run the following query:
SELECT a.relid,
a.relowner
FROM iirelation a,
iitables b
WHERE a.relowner != '$ingres'
AND a.relpgtype < 64
AND b.table_type = 'T'
AND b.table_reltid = a.reltid
AND b.table_reltidx = a.reltidx;
The first time the DBMS Server attempts to access the affected table, a warning message W_DM9C94_TBL_OLD_FORMAT is written to errlog.log. On an Ingres 11 cluster instance, the affected table will be treated as read only until you run a MODIFY statement or a "usermod dbname -uowner table_name" command. On a non-cluster Ingres instance, although transactions can be continued on the table, we recommend running the same MODIFY statement or usermod command.
Instead of using multiple MODIFY statements or issuing the usermod command for each user, you can rebuild all the affected tables in a database by issuing the following command:
upgradedb dbname -modify
• The error message W_DM9C94_TBL_OLD_FORMAT may be seen when tables are using the old page format (before Y2K38 support was introduced in Actian X 11.0). The new page format means that tables that are using the 2K page size may not be able to hold the same volume of data and may fail with E_DM0138. The solution is to change the page size of the table to a larger page size. For details, see Actian Knowledge Base Document “Loading Data into 2K Table May Fail with E_DM0138.”
• Before upgrading to Ingres 11, we recommend setting the DBMS configuration parameter "blob_etab_page_size" to 8K page size if its current value is 2048 or 0. Doing so will prevent the "upgradedb -modify" operation failing to rebuild tables created prior to the upgrade. The tables affected by this error use 2K page size and contain long varchar (BLOB) columns. The upgradedb operation reports error E_QE001D, and errlog.log reports errors E_DM000E and E_QE001D. Because 2K page size is no longer a valid option for an extended table in Ingres 11, the extended table will be rebuilt using the value of the parameter blob_etab_page_size specified in config.dat.
Ingres 10.0
After upgrading from a pre-2.6 release to 10.0, you must re-modify your B-tree tables.
Last modified date: 01/30/2023