IBM DB2* Universal Database Text Extender Version 7.1 for Sun Solaris** F i x P a k 4 -------------------------------------------------------------------------------- Content 1. Prerequisites 2. Changes 2.1 Enhancements 2.2 Fixed problems 3. Hints 4. Current restrictions 5. Installation Extender indexes 5.1 Sun Solaris 5.1.1 Installation instructions 5.1.2 Updating the instances 5.1.3 Recreating the Text Extender indexes -------------------------------------------------------------------------------- ================================================================================ 1. Prerequisites ================================================================================ DB2 Text Extender Version 7.1 must be installed on the machine before you install the FixPak. To install FixPak 4, you need 31 MB (OS/2 and Windows NT) 90 MB (Unix) free disk space on the target drive. ================================================================================ 2. Changes ================================================================================ -------------------------------------------------------------------------------- 2.1 Enhancements -------------------------------------------------------------------------------- 1. New languages are supported. For a complete list see the DB2 UDB Text Extender Administration and Programming Guide V7.1.0.1 page 43f 2. New server administration command file txservice Use this command file only in case of Text Extender problems. The command file collects information about your environment and writes the information to a file. You can review the file and send it to IBM service. 3. Limit of 60 sections to be searched for has been removed. 4. Limit of 32KB for a paragraph in an XML document. 5. New option for the server ini file 'dessrv.ini' to specify that signal handling is switched off to force a core dump. -------------------------------------------------------------------------------- 2.2 Fixed problem -------------------------------------------------------------------------------- 1. Various problems with the sample GUI 'txwizard'. 2. Invalid error message RC_SE_CCS_NOT_SUPPORTED if the thesaurus path name was too long. 3. Problem that the # character could not be searched. 4. Problem that the last word of a document could not be searched. 5. Missing XML support for SUN. 6. G_LADN_ABEND problem. The fix for this problem requires to reindex the data. 7. Signal handling setting for SIGPIPE is preserved now. 8. Various documentation updates. See new version of DB2 Text Extender Administration and Programming Guide V7.1.0.1 which is available for download from the FixPak directory. 9. Problem that Text Extender has overwritten DB2 signal handlers. 10. Search for national language characters (e.g. German Umlaut) in a UTF8 database. 11. RC 201 problem when enabling a UTF8 database with "db2tx enable database" if running in a DB2 V7.2 environment. 12. Fix for search crash using front and end masking (e.g. %term%) 13. Search crash when search_result UDF is used and not all columns of the temporary table have been selected. 14. Semaphore timeout problem when running index updates and searches in parallel. 15. Thesaurus search for accented characters did not return a result. 16. RC_NOT_ENOUGH_MEMORY for large result lists. ================================================================================ 3. Hints ================================================================================ 1. UNIX: For databases with CCSID's other than 850 the sample command file 'txsample' has to be adapted. Change the CCSID value in the ENABLE TEXT COLUMN commands for the NGRAM indextype from 850 to your database CCSID. 2. If the desdiag.log file shows an error ...Shared Memory Access Failed o UNIX Try to remove pending shared resources using the new command txipclean Refer to the Administration and Programming Guide for more information. o Windows Stop and start Text Extender. o If you encounter that only one node has been indexed correctly in a EEE environment, check the the system time is the same on all nodes. 3. If you use the utility db2look to generate ddl statements for a Text Extender enabled database, check if the statements to create the Text Extender distinct types are defined with the FOR BIT DATA option. e.g. create distinct type db2tx.DB2TextH as VARCHAR (60) FOR BIT DATA with comparisons; db2look generates the distinct types without this option. This causes all handle values that are generated by Text Extender to be corrupted. Either change all these statements manually or copy the statements stored in the Text Extender installation directory ddl, file descudt.ddl You can also run this file with : db2 -tvf descudt.ddl from the operating system command prompt. ================================================================================ 4. Current restrictions ================================================================================ o To run Text Extender administration commands from a Windows NT or Windows 2000, the setting for the DB2 service in the Services window of the Control Panel is set to a valid administrator userid. o For Windows NT, the environment variable DB2INSTANCE must be set as system environment variable. If it is not set, the utilities txstart, txstatus and txstop cannot be called. o STEMMED FORM OF for indextype normalized is only supported for US_ENGLISH. o The REFINE UDF does not support IS ABOUT or THESAURUS queries as a query refinement. o Masking is not allowed for SOUNDS LIKE. o The maximum number of model names in the model definition file is 50. o Stemming is not supported for thesaurus searches on a linguistic indextype. o Search on a UTF8 database is currently not supported. o The backslash cannot be used as escape character for CCSID 954 o The following formats are not supported for DBCS languages: AMI, FFT, MSWORD, RFT, RTF, WP5 o Values for attribute sections for search must be SBCS characters. o Running multiple enabling commands in a EEE environment on the same table in parallel may result in RC 507 shown with db2tx get index status. Workaround: Run the enable commands sequentially. o For Windows 2000 the system and user locale must match. o Linguistic search requires correct spelling with respect to the first character regarding upper/lowercase if the search term requires linguistic normalization. E.g. German "Edelmetalle" in the document is not found if the search term is "edelmetalle". The reason is that at indexing time the plural "Edelmetalle" is normalized to singular "Edelmetall" and then to "edelmetall" which is stored in the index. A search for "edelmetalle" is processed by a dictionary lookup. As the dictionary contains the correct German spelling for a noun "Edelmetalle" the search term is not found. In this case the search term is used for the index lookup. This results in a "not found" condition. o Word decomposition. Word decomposition is only possible if the parts of a compound word are available in the base form. E.g. German "Kindersprache" is stored in the index as "kindersprache" and "sprache". The index does not contain "kinder" or "kind" as "Kinder" in "Kindersprache" is the plural of "Kind". ================================================================================ 5. Installation ================================================================================ General steps for all platforms: Perform the following steps for each instance: (You can get a list of all Text Extender instances by issuing the txilist command as described in the Text Extender documentation): 1. Terminate all Text Extender Command Line Processor sessions. 2. Make sure that the Text Extender server has been stopped by issuing : txstop 3. Make sure that all DB2 connections are terminated by issuing : db2 terminate 4. Check if applications are still active by issuing: db2 list applications show detail 5. If applications are still running, check that you can stop running DB2 applications and enter the following command: db2 force applications all -------------------------------------------------------------------------------- 5.1 Sun Solaris -------------------------------------------------------------------------------- To install the FixPak on your machine, there are three major steps to perform: 1. Installing the Text Extender FixPak software (chapter 5.1.1) 2. Updating the Text Extender instances (chapter 5.1.2) 3. Recreating the Text Extender indexes (chapter 5.1.3) In a DB2 EEE environment you have to perform step 1 on all physical nodes. Steps 2 and 3 have to be performed for the master node only. The steps are described in the following chapters. 5.1.1 Installation instructions Perform the following steps to install the FixPak: 1. Download the fixpack file in FP4_TE_U487312.tar.Z in binary format. 2. Uncompress the file using the following command: uncompress FP4_TE_U487312.tar.Z 3. Untar the uncompressed file by issuing tar -xvf FP4_TE_U487312.tar 4. Important: Perform the general steps at the beginning of chapter 5. 5. Install all filesets from the FixPak that were previously installed on your machine by using the graphical 'admintool' or the pkgadd command of the operating system. 5.1.2 Updating the instances Enter the following command for all Text Extender instances on your system (You can get a list of all Text Extender instances by issuing the command /opt/IBMdb2tx/V7.1/instance/txilist with root authority): /opt/IBMdb2tx/V7.1/instance/txiupdt my_TE_instance my_DB2_instance where my_TE_instance is the instance owning user-id of your Text Extender instance and my_DB2_instance is the instance owning user-id of the related DB2 instance, which is usually the same user-id as for the Text Extender. 5.1.3 Recreating the Text Extender indexes Because the structure of the Text Extender indexes changed slightly with the current FixPak, you have to drop and recreate it before you can use it with the new code. See the description of the Text Extender commands DISABLE TEXT COLUMN/TABLE/FILES and ENABLE TEXT COLUMN/TABLE/FILES in the Text Extender documentation on how to drop and create Text Extender indexes. -------------------------------------------------------------------------------- For any problems related to that FixPak, please contact: IBM Germany Content Management Service and Support Dept: 5248, Schoenaicher Str.220, 71003 Boeblingen Phone: +49-7031-16-4254 E-Mail : met@de.ibm.com -------------------------------------------------------------------------------- * are trademarks or registered trademarks of IBM Corporation. ** are trademarks or registered trademarks of their respective owners. -------------------------------------------------------------------------------- ******************************************************************************** ** ** (c) COPYRIGHT INTERNATIONAL BUSINESS MACHINES CORPORATION 2000. ** ALL RIGHTS RESERVED. ** ********************************************************************************