MDM XU ReleaseNotes

download MDM XU ReleaseNotes

of 22

description

sdsd

Transcript of MDM XU ReleaseNotes

  • Patch B Release Notes Whats New in XU Rollup Patch B Upgrading from XU Rollup Patch A Issues Resolved in XU Rollup Patch B Known Issues in Siperian Hub XU Rollup Patch B (and XU Patch A, GA)Patch A Release Notes Known Issues in Siperian Hub XU Rollup Patch A (and GA) Issues Resolved in XU Rollup Patch ASiperian Hub XU Release Notes Whats New Before Installing Hub Store Hub Server Hub Cleanse Match Hub Console Troubleshooting Tips Resolved Issues and Enhancements Known Issues

    Siperian Hub XU Rollup Patch B Release NotesSiperian Hub XU Rollup Patch B is a hotfix rollup release. These release notes contain last minute product information andupdates. All previous Siperian Hub release fixes are rolled into this release.Note: ** indicates issues that were in earlier releases that have been resolved in the Siperian Hub XU Patch Brelease.

    Whats New in XU Rollup Patch BSupport for these features: Backup and Recovery WebSphere Clustering Oracle RAC

    The rollup also includes: Transactional multi-merge Multi-merge performance improvement Support for AddressDoctor (Platon) data files from December, 2007 or later with the version of the AddressDoctor engine (with

    Platon build 4.1.11.384). Support for Trillium Director 7.8 Support for Group One CDQ 5 All hot fixes from the previous releases have been rolled into this patch. Please review the resolved issues in XU Rollup Patch B andthe issues from earlier releases that are resolved in Siperian Hub XU Rollup Patch B (indicated with **).Please contact SiperianSupport for additional information about the new features in this rollup. Important: New hot fixes for XU will be built only on top of XU Rollup Patch B.

    Upgrading from XU Rollup Patch AYou can upgrade to this patch from XU Rollup Patch A, XU GA or any patched version of XTSP2, for example, XTSP2 Patch A orXTSP2 Patch B. The patch install and the ORS upgrade process is exactly the same as our regular patch upgrades. Please reviewthe Install Guide for detailed instructions about how to upgrade to this patch or do a full install. General steps to upgrade to XU Patch B from XU Patch A (or an earlier patched version) are: Upgrade the MRM Server using mrm server install command: hub_install.exe Upgrade the MRM Cleanse using mrm cleanse-match install command: hub_cleanse_install.exe Upgrade the Master using the command: sip_ant updatemasterdatabase Upgrade the ORS using the command: sip_ant updateorsdatabase If you use Platon AddressDoctor for address cleansing, you need to upgrade to AddressDoctor build 4.1.11.384 and update

    your AddressDoctor data files to December 2007 or later. Please review the Install Guide for detailed instructions about how

    Siperian Hub XU Rollup Patch A Release Notes

    1 of 22

  • to upgrade Platon AddressDoctor, or contact Siperian Support . Please contact Siperian Support for more information about the Group One CDQ V5 Adaptor.

    Please note that there is no change to the Master database in this patch since XU Rollup Patch A. Several issues reported by Customers on Oracle 10.2.0.1 and Oracle 10.2.0.2 were resolved by upgrading to Oracle 10.2.0.3.

    For Oracle versions, our recommendation is to use Oracle 10.2.0.3. You should use JDBC driver 10.2.0.3 with Oracle10.2.0.3 for all app servers, except IBM WebSphere.Note: The Oracle JDBC driver 10.2.0.3 is incompatible with IBM WebSphere JRE 1.4.2 on Windows. You need to use theOracle JDBC driver 10.2.0.2 instead when using an Oracle 10.2.0.3 database with WebSphere.

    If your ORS is not upgraded to XU Rollup Patch B, you will get the following warnings when you run MET validation against the ORSand the schema will also fail to load in the console.2 Warning SIP-PV-10108 System column 'GLOBAL_NOLOGGING_IND' in table 'C_REPAR_DB_RELEASE' is not found.3 Warning SIP-PV-10108 System column 'GLOBAL_NOLOGGING_IND' in table 'C_REPOS_DB_RELEASE' is not found. The javadocs in this XU Rollup Patch B are not up to date. Please contact Siperian Support to get the latest resource kitsoftware for the updated javadocs.

    Using User ExitsNote: Applying the new cmxue packages will overwrite any existing user exits; you would need to backup and re-apply any customcode.The range of error codes for user exit errors is -24300 to -24399. Do not raise an exception within the user exit. Each of the userexits has a return code and error message out parameter. Set these to indicate an error. Back to top

    Issues resolved in XU Rollup Patch BThese issues have been resolved in XU Rollup Patch B. See Known Issues in Siperian Hub XU Rollup Patch B and Known Issuesin Siperian Hub XU Rollup Patch A for a list of issues that have not been resolved. Note that some cases there are new behavior toresolve issues. The old and new behaviors are noted below. (Siperian uses an internal bug tracking system with 5 digit numbers assigned to each PIR (request). 4 digit numbers representsupport case IDs, 5 digit numbers internal requests.)

    Cleanse Engines:

    AddressDoctor17353: Support for AddressDoctor (Platon) December-2007 or later dataset.Note: You must use AddressDoctor data files from December 2007 or later with the version of the AddressDoctor engine (withPlaton build 4.1.11.384) distributed in XU Rollup Patch B. Group One16452: Support for Group One CDQ Version 5.Note: We no longer support Group One Enterprise V2.7. Please contact Siperian Support for more information about the Group OneCDQ V5 Adaptor. Please note the Group One chapters in the Windows and UNIX Install Guides have been updated for Patch B withinformation for the Group One CDQ Adaptor. Please note the Group One chapters in the Windows and UNIX Install Guides havebeen updated for Patch B with information for the Group One CDQ Adaptor. Trillium Director17945: Support for Trillium director 7.8.Note: According to Trillium, there are no major changes between 7.8 and 7.7b. Users will only have to replace the Trillium 7.7binstall with Trillium 7.8.. No build/property changes are required to use Trillium 7.8. Staging:16861: Drop ORAW temp tables created during the staging process.17546, 16861: Inconsistencies in PRL and ORAW tables when delta detection and raw record retention are enabled.17862: Bad query plan on insert into staging table during raw retention by adding hash join hint. Load17486: Load by rowid_object fails with error "Load_master successful with missing lookup records / invalid rowid objects" for rowidsthat are merged.

    Siperian Hub XU Rollup Patch A Release Notes

    2 of 22

  • 17802: Invalid lookups are not rejected when "allow null foreign key" option enabled. This fix changes the following behaviorOld behavior: If you have Allow Null Fkey enabled: - The valid values will be processed as per the lookup - The NULL values will be processed as NULLs - The invalid values will be processed as NULLs. We do not reject/provide warning for invalid lookup valuesNew behavior: If you have Allow Null Fkey enabled: - The valid values will be processed as per the lookup - The NULL values will be processed as NULLs - The invalid values will be rejectedSee Configuring Operational Record Stores in the Siperian Hub Administrators Guide for more information about thelookup functionality.

    Merge

    17749: Consolidation_ind can be reset to 4 on child when merging parent regardless of whether the child is consolidated ornot if Requeue_On_Parent_merge_ind is set to 2 via the backend. Below is an example of a SQL script to update theRequeue_On_Parent_merge_ind:UPDATE C_REPOS_TABLE SET REQUEUE_ON_PARENT_MERGE_IND = 2 WHERE TABLE_NAME = '';Old behavior: Supports the following two scenarios:

    requeue_on_parent_merge_ind set to 0: does not queue any related children on parent merge. requeue_on_parent_merge_ind set to 1: queues the related children if consolidation_ind is set to 2 only when parent records

    are merged.New behavior: Supports an additional scenario:

    requeue_on_parent_merge_ind set to 2: queues the related children even if consolidation_ind is set to 1 or 2 when parentrecords are merged.

    See About Matching and Consolidating Data in the Siperian Hub Administrators Guide for more information about theconsolidation functionality.

    This is a change in design to the Immutable Rowid Object, the description for the Immutable Rowid Object section has not beenupdated in the Administrators Guide. The description should now say:

    Immutable Rowid Object: Allows you to designate a source system as an immutable source, which means that records from thatsource will beaccepted as unique and, once a record from that source has been fully consolidated, it will not be changedeven inthe event of a merge. Immutable sources are also distinct systems. All records are stored in the Siperian Hub as master records. Forall source records from an immutable source system, the consolidation indicator for Load and PUT is always 1 (consolidated record).If REQUEUE_ON_PARENT_MERGE_IND is set to 2 on the child via the backend, the consolidation indicator will be set to 4for the child record in the event of merging parent.

    Note: When the requeue_on_parent_merge_ind is set to 2, it is not visible in the console. It is shown as unchecked.

    SIF API17583: PUT SIF API call should handle special characters automatically within the SIF call so that caller doesnt have to do itexplicitly.This is a change in behavior.

    Old behavior:- Special characters have to be escapedSpecial characters include \ ~These characters had to escaped in the old behavior like this: \\ \ \~

    New behavior: - Special characters dont need to be escaped before making the API call. However, if you have custom code thatescaped special characters in the past, you would need to update your custom code to factor in the new behavior.

    See the Siperian Hub SIF Guide for more information about the SIF functionality.

    17728: Put SIF API needs to process null values like the CleansePut API, for example when no value is specified for a field, it shouldbe set to null. This is a change in behavior. Old behavior: Put fails when no value is specified to INT, NUMBER or DATE column. For example,

    CONSUMER_ID INT_COL

    Siperian Hub XU Rollup Patch A Release Notes

    3 of 22

  • NUMBER_COL DATE_COL

    New behavior: PUT goes through with NULL value in INT, NUMBER or DATE column.Please note:

    bigIntegerValue, bigDecimalValue, etc. are strongly typed elements. The format used below to set null values in thoseelements is not valid. The format used below is NOT valid:

    INT_COL

    NUMBER_COL

    The format below is the valid way to insert null values for INT, NUMBER, etc.

    INT_COL

    NUMBER_COL

    See the Siperian Hub SIF Guide for more information about the SIF API functionality.

    17857: PUT/CLEANSE_PUT fails to insert a new record with default value in non-nullable varchar column.17887: PUT/CLEANSE_PUT fails with default value in non-nullable date column.17641 tokenize SIF API call fails to close the SSA connections properly.17914: Compatibility issues to make SIF API request through .Net. since .Net has a different way of interpreting optional parametersand null values. Schema manager:17561: Nullable value does not get set on the PRL if landing table is updated resulting in SIP-PV-10210 errors17568: synch up staging column sequence number when the column sequence number changed in BO17570: C_REPOS_CL_COL_PARAM not cleaned up when landing cols are dropped resulting in SIP-MV-11932 errors17743: trust_ind not set on the staging properly if trust is disabled in the base object resulting in SIP-MV-10306 errors Proxy User17772: Auto link job fails when Proxy User is enabled17748: Proxy_Role is missing for BO, VCT and CTL tables when running automerge with automerge_ctas is set to 1 (default is -1). Performance16937: Allow Transactional Multi-merge - multi-merge must treat the entire set of records to be merged as one batch.16936: Performance Issues with PUT inside of Merge service. PUT performance has been optimized to reduce the latency of someof the XREF updates.16492: Further optimize the multi-merge performance by removing the unnecessary analyzing and parallel execution on temp tables.Fixed the bad execution plan of the long running update on hfkm table. HM17271: Indirect search query constructed incorrectly in the SearchHMQuery call17424: Changing Entity Direction in HM configuration causes Java Null Pointer Exception. Environment Staging:17553: The Environment Staging script (11_migration_repos_seq_keycount.sql) was not setting Oracle sequence numbers correctlyand caused constraint violation errors during data load (Load/Put).17591: Sequence on GBID columns do not get created on the target schema17482: Environment staging scripts do not create the STRP tables as IOT tables

    Siperian Hub XU Rollup Patch A Release Notes

    4 of 22

  • 18019: The Environment staging script 09_check_repository_vs_physical.sql ignores UTF8 settings when generating ALTERstatements for modified columns. MET17250, 17036: Fix SIP-MV-10218 Validation warnings that are reported incorrectly against the RMT, VOK and VPC columns on theVCT, VXR and related History tables. Similarly, fix SIP-MV-10218 validation warnings that are reported incorrectly against theShadow key columns (S_*) on the XREF tables.17432: Metadata Manager displays incorrect warning messages indicating that columns are not used.17731: Update the SQL consistency check for Packages to address erroneous SIP-PV-10703 errors17712: Remove duplicate shadow column check to address erroneous SIP-MV-10201 errors Resource Kit:17779: Updated the Deployment panel to clearly state that the user cannot use our deployment process if they choose not to deploythe resource kit sample apps. Others:17610: Inconsistency - STRICT_IND is nullable in the hub metadata but not nullable in the database17970: The datasource connection from WebSphere may give a connection failure error on the Oracle RAC. This is not a Siperianissue; you will need to use the actual node name for the SID instead of the Oracle RAC instance name in the datasource connectionproperties. Back to top

    Known Issues in Siperian Hub XU Rollup Patch B (and XU Patch A, GA):Listed below are additional issues in XU Rollup Patch B (and in XU Rollup Patch A, XU GA releases).

    Cleanse Engines:

    AddressDoctor18010: AddressDoctor (Platon) might error out when cleansing large datasets because of a memory Leak. Please contact SiperianSupport for an update on this issue. Data Manager:17268: Data Manager allows the creation of Child records even if the foreign key is not specified.17918: You might run into data access exception in Data Manager to view LINK style Base Objects using setup Administrationbutton.Workaround: Simply ignore the error and close the popup windows. Documentation:18015: Only the Install guides have been updated to indicate the XU Patch B version, all the other documentation guideslike the User Guide, Admin Guide, and SIF Guide still say Siperian Hub XU on the version page.The Group 1 Enterprise Server cleanse engine chapters in the UNIX and Windows Install Guides have been updated withGroup 1 CDQ 5 information.17541: Javadocs are not up to date. Please get the latest resource kit software from Support for the updated javadocs.

    Stage:17770: Staging job fails when landing table has not nullable columns (varchar) with default value Load:17569: Load by rowid object rejects rows not found in XREF and HMRG table but doesn't populate %_REJ table if the landing tableis truncated.17829: Endless loop in LOAD_MASTER standard_abort exception handlerWorkaround: Run the metadata validation tool and manually recreate any missing constraints to avoid the error condition that leadsto this endless loop. Contact Siperian Support if you need more information or assistance.18048: Records with null or invalid foreign keys will not get rejected if the relationship is based on unique keys and the lookup isbased on the primary Key of the XREF table. Users will only get a NO ACTION in such cases. These records will also not be insertedinto the Base Object. Match:17660: External Match result do not appear in Batch Group Logs Viewer17548: The ruleset filter may cause an error to be thrown if there are records set to consolidation_ind=3 at the time of Match Jobexecution.Workaround: Reset the Base Object records to 4 and re-execute the Match Job.

    Siperian Hub XU Rollup Patch A Release Notes

    5 of 22

  • 17620: The Link Style MFD/PK_Match job may return a warning if you execute these jobs after running clean_table.17663: The Link Style match job does not use the BVTB table. Merge:17890: _TMST temp tables are left behind after automerge process and automerge_ctas_ratio=1.Workaround: manually drop these temp tables17912: Temp tables are left behind after multi-merge call.Workaround: This is by design; the caller should manually drop the temp tables. Unmerge:16154: Unmerge will not cascade to child records properly if the Child is loaded after parent merged. This is because the ChildHFKM table does not contain the metadata that links the Child records to the original Parent record. (Known issue from XU Patch A) Merge Manager:17814: "Accept All Records without a Match" affects records that aren't queued for merge. API:17684: API SearchQuery issue - need to pad parameter values for ROWID_OBJECT columns HM:17795: Load fails if HM relationship object columns are called rowid_bo1, bo2

    Workaround: drop and recreate the relationship object, for example don't use rowid_bo1, rowid_bo2 as column names whenre-creating the relationship object.

    17490: HM does not use security on ProfilesWith the introduction of SAM in XU, there are no changes to the security functionality in any of the Data Steward tools (DataManager, Merge Manager, Hierarchy Manager). These tools generally don't use the APIs where security is enforced. Theymake limited use of the security assignments:- Data/Merge Manager - list of packages is limited by security roles. Ability to Merge is not constrained by security. Put isthe only API used - security is enforced there.- Hierarchy Manager - list of profiles is limited by security roles. A user has access to a profile that he has any privileges on.

    17499: New sandbox validation fails if "Validate Data" is checked17384: The console does not stop the user from dropping columns from HM packages through the Package Manager. This maycause issues during validation and in the HM UI later. In general, modifications to HM packages (for example dropping requiredcolumns from the package) can result in the HM Package and also the HM unusable.17515: NPE calling updateRelationship with invalid Hierarchy as part of record tag MET Validation17437: The metadata validation tool may report data length inconsistencies on system tables when run against a UTF8 database.These warnings can be safely ignored.17889: The MET tool does not report conditions that do not have outputs in the cleanse function UI. This will only happen if youmapped the condition previously and then deleted the output from the condition in the cleanse functions UI.The out put from MET will indicate a Fatal error in this case. Please contact Siperian Support if you get a Fatal error when validatingyour ORS.1 Fatal Error SIP-PV-10000 Unable to load Siperian metadata. Others:17798: The ORS group name may disappear from the Security Workbench if the ORS group and user share the same name.Workaround: Avoid using the same name for the User and the ORS Group.17910: The Mapping UI/cleanse functions UI can fail to load if you have a mapping that has conditions without any outputs. This willonly happen if you mapped the condition previously and then deleted the output from the condition in the cleanse functions UI.Workaround: drop and recreate this condition.17797: Typo in creating a proxy user in Windows Hub installation guide.17947: The Resource Kit is supported only on Windows. Please review the Resource Kit User Guide for more information about thesupported platforms. Back to top

    Siperian Hub XU Rollup Patch A Release NotesSiperian Hub XU Rollup Patch A is a hotfix rollup release. These release notes contain last minute product information andupdates. All previous Siperian Hub release fixes are rolled into this release.

    Known Issues in Siperian Hub XU Rollup Patch A (and GA):All known issues in the XU GA release notes apply to the XU Rollup Patch A. Listed below are additional issues in XU Rollup Patch A

    Siperian Hub XU Rollup Patch A Release Notes

    6 of 22

  • (and GA). ** Platon17353: The AddressDoctor (Platon) December-2007 dataset is not supported in this Rollup Patch. You would need to use theSeptember Data files from Address Doctor with Platon Build 334. ** Hierarchy Manager**17134: Cannot drop HM entity type object table because of constraints to HM node.Workaround: Do not use the Remove option in the impact analysis screen to drop HM objects. Instead revert the HM objects toregular base objects and then use the drop option from the Schema Manager. MET Validation17254: The MET Validation tool reports the SIP-MV-11903 warning incorrectly if you have HM Foreign Key Objects (for e.g.SIP-MV-11903 HM package 'PCK_PROD_READ' does not contain all records from the underlying table)17253: The MET Validation tool reports the SIP-MV-11931 warning incorrectly if you have HM Foreign Key packages (for e.g.SIP-MV-11931 Required column 'REL_START_DATE' is not found in the package 'PCK_PROD_WRITE' of profile 'Default'relationship type 'Manufactures')** 17250, 17036: SIP-MV-10218 Validation warnings are reported incorrectly against the RMT, VOK and VPC columns on the VCT ,VXR and related History tables even if trust is setup for the columns.Similarly, SIP-MV-10218 validation warnings are reported incorrectly against the Shadow key columns (S_*) on the XREF tables17265: The Validation tool does not report incorrect ROWID_PARENT values in History tables (SIP-MV-10219 )17350: Metadata tool does not report views/packages that are invalid17327: The Metadata tool does not report missing defaults on RAW/OPL/PRL Merge** 16937: Multi-merge must treat the entire set of records to be merged as one batch.17125: History Viewer - Merge History Panel does not display the correct time. Migration16918, 17082: XU migration might take longer to process an update SQL in "2007050902_ldc_upd_tree_unmerge_struct_data.sql". Oracle:17398: The Oracle JDBC driver 10.2.0.3 is incompatible with IBM WebSphere JRE 1.4.2 on Windows. You need to use Oracle JDBCdriver 10.2.0.2 instead when using an Oracle 10.2.0.3 database with WebSphere. For more information, see: http://www.ibm.com/developerworks/forums/dw_thread.jsp?forum=367&thread=172043&message=13980613&cat=10&q=IBM+JDK+CERTIFICATON%3A+THIN+DRV+TEST+GOT+BAD+CONNECTIONS#13980613 ** Performance16936: Performance Issues with PUT inside of Merge service. SIF API12786: incorrect API privilege settings, clearAssignedUnmergedRecords API has READ privilege but it should have WRITE and theunmerge API has WRITE but it should have MERGE17120: describeSiperianObject API does not support BATCH_GROUP type.17123: validateMetadata API does not return message in the response to indicate if the request is successful.17087: The registerUsers API call does not set the default database for the new user, which causes a null pointer exception whentrying to login as that user.17039: The SIF APIs fail if you provide the IP address in the hostname.Workaround: Create an alias in the hosts files on Windows client and Solaris server.16873: Records set as ON_HOLD by setRecordState API call cannot be seen from On Hold screen in Merge Manager.17097: describeSiperianObject API for MATCH_RULE_SET and MATCH_COLUMN can be executed without privilege.17110: describeSiperianObject API for HM-SANDBOX can be executed without privilege.17121: HM APIs do not use the configurable column names.17177: SearchMatchRequest API fails with error SIP-14105: Unexpected error processing request if a package is defined on anXREF table instead of BO.Workaround: Change the packages to be based on BO instead of XREF.17184: SIF API cleanse call only works with EXECUTE privilege on BO mapping.17215: getLookupValue API fails if Package UID & BO columns have only READ privileges.17214: acceptUnmatchedRecordsAsUnique API can be executed with only BO UPDATE privilege.17203: getUnmergedRecordCount API can be executed with only BO READ privilege.17269: getBatchGroupStatus API returns incorrect runStatus when batchGroup fails.17271: search query constructed incorrectly by the SearchHMQuery API. The API does not substitute ROWID_OBJECTs and codecolumns correctly based on the Base Object settings. Unmerge

    Siperian Hub XU Rollup Patch A Release Notes

    7 of 22

  • 16154: Unmerge will not cascade to child records properly if the Child is loaded after parent merged. This is because the ChildHFKM table does not contain the metadata that links the Child records to the original Parent record.11981: Concurrent Unmerge on the same rowid_object is not supported.15761: The Rowid_Object returned by Unmerge is not correct if you unmerge a subtree from the Data Manager. The storedprocedure returns the rowid_object correctly; the issue is in the GUI. Other16182: Custom buttons allow you to define buttons/functions that can be associated with any or all of Data, Merge, and HierarchyManagers. However, the code that fetches the functions for a specific tool always gets only the functions that are targeted for DataManager.**16540: You might run into .Net (Visual Basic) compatibility issues when using Visual Basic applications.17052: Integrity constraint violation when add/update privileges in Role if only Mappings is set to SECURE when in the SECURERESOURCE UI.16739: NullPointerException from View that is built on top of C_REPOS_AUDIT in Query Manager.17105: ORA-06502: PL/SQL: numeric or value error: associative array shape is not consistent with session parameters in the debuglog. This is an Oracle issue that has been fixed in Oracle 10.2.0.4. as stated in metalink Doc ID: Note:467688.116384: Old raw records might not be removed as expected when the retention period is reduced to certain days.16307: The console currently allows Anti-match (not equal match) and null matches non-null as coexisting options. This is an invalidcombination.**17363: Help Text on the Patch install screens refers to upgrading to XU GA instead of the XU Rollup Patch A. (The patch installerfor XU Patch B has been updated.) Back to top

    Issues resolved in XU Rollup Patch A(Siperian uses an internal bug tracking system with 5 digit numbers assigned to each PIR (request). 4 digit numbers representsupport case IDs, 5 digit numbers internal requests.) 14592: Allow proxy settings in DnB adapter. This issue was resolved in XU GA but the procedure was not documented.

    Users would need to add the following possible properties in the cmxcleanse.properties file to allow DnB to route via aproxy server

    cleanse.library.dnb.property.proxy.server=cleanse.library.dnb.property.proxy.port=cleanse.library.dnb.property.proxy.userid=cleanse.library.dnb.property.proxy.password=

    17012: The environment staging doc should include information about users and package assignments.16347: Optimizations have been made to multi-merge performance.16365: Optimizations have been made in the Automerge batch process to improve performance.16368: Error running Stage job because of memory leak in Platon. Please review the Install Guide for detailed instructions about howto upgrade Platon.16376: Ability to rollback if there are failures when executing the multi-merge API.16428: Large number of columns causes Load to fail during the creation of the _TLL table with error: ORA-06502: PL/SQL: numericor value error: character string buffer too small.16433: Load update does not fully populate history tables.16499: Environment Staging scripts need to migrate link tables and create sequences for them.16544: Automerge job fails with a breakpoint error when parsing GM_COL_LIST if the system_id (machine_id) doesnt start with Sand there are a large number of systems and trust/validation enabled columns.16573: Load by rowid_object job loops without any error message if the system_id (machine_id) doesnt start with S and there area large number of systems and trust/validation enabled columns.16691: Stage job with a large number of columns fails on creation of _DLT table with error: ORA-06502: PL/SQL: numeric or valueerror: character string buffer too small.16707: Cleanse_Put and Put fails with "character string buffer too small" error when the field value exceeds 2000 characters.16723: NullPointerException when getting into HM configuration tool if the relationship type is not associated with hierarchy.16772: The performance issue that occurred during Incremental load-when archive-log mode is turned-on and you have a largenumber of trust/validation columns has been fixed.16775: The performance issue in Automerge after incremental load if you have a large number of trust/validation columns has beenresolved.16883: The direction of the parent to child relationship is switched incorrectly when configuring the relationship between party toparty in HM.16886: AM deployment fails on JBoss.16899: POST_LOAD is not called when every record in staging is rejected.

    Siperian Hub XU Rollup Patch A Release Notes

    8 of 22

  • 16904: Foreign Key constraint might get dropped from the base object after initial data load if CTAS is on.16911: Refresh in Batch Viewer causes a NullPointerException if the batch operation is still in progress.16942: The base object physical table may sometimes disappear when running match job if local / remote cleanse-match servers areset off-line/on-line.16974: MRM creates Message Queue connection without any credentials.

    This is resolved in XU Rollup Patch A. To use this capability, you would need to add the following properties in\hub\server\resources\cmxserver.properties: [jms connection factory jndi].qcf.username=[username] [jms connection factory jndi].qcf.password=[encrypted password] AND/OR [jms connection factory jndi].ctx.username=[username] [jms connection factory jndi].ctx.password=[encrypted password] Note: 1. Password must be encrypted using blowfish 2. To encrypt a password, open cmd and cd C:\siperian\hub\server\lib\: java -cp siperian-common.jar com.siperian.common.security.Blowfish password

    16978: Manual merge is slow when a large amount of data needs to be updated in _HMXR table.17011: The Prerequisites for Running the Migration scripts section in the environment staging doc has been updated to includeinformation about the C_REPOS_ERROR_MSG table.17033: Unique constraint violations might happen when creating mappings after deletes of base object/staging tables.17196: ExecuteBatchGroup Stored Procedure does not accept batchGroupID. The get_batchgroup_status, execute_batchgroup andreset_batchgroup stored procs have been updated to accept batchgroupUID now.17219: NullPointerException when non-admin user with read-execute privileges adds/updates entities in HM.17226: Error reverting HM BO when trust is enabled.17299: Out of Memory errors when executing cleanse against WebSphere. This may happen because the 9i Oracle helper class isused in the data source by default.Workaround: WebSphere customers should make sure that the right helper class is being used in the datasource. For example, ifyou are executing against a 10g database, use the 10g oracle helper class in the datasource.17331: add SHOW_VERSION to cmxdb package17338: special characters cause issues in cleanse put call Back to top

    Siperian Hub XU Release NotesSiperian Hub XU is a major release. These release notes contain last minute product information and updates to the Siperian HubXU documentation. All previous Siperian Hub release fixes are rolled into this release. Siperian Hub XU includes:

    Master Reference Manager (MRM) The foundation and core product that consolidates multiple customer profiles tobetter identify customers uniquely across all channels, thereby increasing the accuracy and reliability of customer referencedata to all systems.

    Optionally, you can separately license either or both of the following products:

    Hierarchy Manager (HM) Delivers reliable and consolidated customer relationship views that enable organizations tonavigate, analyze, and manage relationships across multiple hierarchies from disparate sources

    Activity Manager (AM) Evaluates data events, synchronizes master data, and delivers unified views of reference andactivity data from disparate sources

    These release notes cover the following topics:

    Whats New Before Installing Troubleshooting Tips

    Resolved Issues and Enhancements Known Issues

    Back to top

    Whats NewFor Siperian MRM

    Metadata Validation Security Models and Management Workflow Integration Reporting and Dashboards

    Siperian Hub XU Rollup Patch A Release Notes

    9 of 22

  • Cleanse Adaptors Match, Link, and Merge Enhancements Performance and Scalability Hierarchy Management Batch Groups Schema Viewer Installer Enhancements New Environments Sample Schema Full support for Oracle RAC 10gR2

    For Siperian HM Foreign Key Support Creation of Entity and Relationship Base Objects within the HM Configuration Tool Creation of MRM packages within the HM Configuration Tool Modification of existing Base Objects from MRM to HM tables

    For Siperian AM

    AM rules on batch events AM Rules API

    Siperian Hub enhancements that affect all products

    Enhancements and additions to the core Siperian APIs API extensions to support full external data entry portal SIF SDK 2.0 Templates that enable users to:

    o Use templates to create your own Business serviceso Create complex interactions such as CPMM (Cleanse Put Match Merge) and CMP (Cleanse Match Put)o Create external definitions of Business Objects

    To learn more about new features and enhancements in Siperian Hub XU, see Whats New document found in the documentationfolder. Back to top

    Before InstallingInstallation requires that you acquire a license file from Siperian, so that the application will work properly. If you do not have the file,you can continue with the install, but the application will not function until the information is provided. Contact Siperian support formore information. Note: Emulated environments such as VMware or Citrix are not supported under any configuration.

    Hub StoreHub Store requires the following hardware and software: Hardware RecommendedProcessor (Intel) 2 CPUs @ 2.0 GHz Pentium 4 or equivalent *Memory 2 GigabytesHard Disk Space Based on customer dataCD ROM Yes* The recommended minimum number of CPUs and minimum memory values are dependent on your data. One of the following operating systems Supported versionsWindows 2003, XP sp2HP-UX 11 11.11.0509. Only for WebSphere 6, you must apply the linker patch PHSS_33035.Solaris 9 5.9 Generic_118558-19 sub4u sparcSolaris 10 5.10 Generic_ 118833-03Linux SUSE 9 SP3, Red Hat Enterprise Linux AS

    v4.3 (2.6.9-34.0.1.ELsmp)AIX 5.3 (with the IBM JDK v 1.4.2)

    Patch 5300-03 Software Supported versions

    Siperian Hub XU Rollup Patch A Release Notes

    10 of 22

  • Oracle 9i (9.2.0.8), 10.2.0.3 - with database server and network client and SQL*Plus(same release). For Oracle 9i, 10.2.0.x use the 10.2.0.2 driver. Several issues reportedby Customers on Oracle 10.2.0.1 and Oracle 10.2.0.2 were resolved by upgrading toOracle 10.2.0.3. You should use JDBC driver 10.2.0.3 with Oracle 10.2.0.3 for allappservers, except IBM WebSphere.Note: Oracle JDBC driver 10.2.0.3 is incompatible with IBM WebSphere JRE 1.4.2 onWindows. You need to use Oracle JDBC driver 10.2.0.2 instead when using an Oracle10.2.0.3 database with WebSphere.

    Hub ServerHub Server requires the following hardware and software: Hardware RecommendedProcessor 2 CPU @ 2.0 GHz Pentium 4 or equivalent *Memory 2 Gigabytes *Hard Disk Space 300 MBCD ROM Yes* The recommended minimum number of CPUs and minimum memory values are dependent on your data. One of the following operating systems Supported versionsWindows 2003, XP sp2HP-UX 11 11.11.0509. Only for WebSphere 6, you must apply the linker patch PHSS_33035.Solaris 9 5.9 Generic_118558-19 sub4u SPARCSolaris 10 5.10 Generic_118833-03Linux SUSE 9 SP3, Red Hat Enterprise Linux AS 32 or 64-bit

    v4.3 (2.6.9-34.0.1.ELsmp)AIX 5.3 (with the IBM JDK v 1.4.2)

    Patch 5300-03 One of the following application servers Supported versionsJBoss 4.0.3 SP1 NOTE: JBoss is not supported on AIX platforms.WebLogic 8.1.6 (Enterprise Edition required), 9.2WebSphere 5.1.1.6, 6.1.0.3 (These are the only versions that are supported. Please contact Siperian Support for more information.) Software SupportedJDK 1.4.2_13 or 1.5 Note: Clustering is supported only on the WebLogic application server.(**On XU Patch B and above, clustering is also supported on the WebSphere application server.) The Oracle client must be installed on every machine on which you wish to use Hub Server. This Oracle client requires Oracle Javautilities, so you must install these utilities on each machine on which you have Hub Server as well. The Oracle Client and OracleJava utilities are part of your Oracle distribution.You must install SQL*Loader on each machine on which you use Hub Server. SQL*Loader is part of your Oracle distribution.

    Hub Cleanse MatchHub Cleanse Match requires the following hardware and software: Hardware RecommendedProcessor (Intel) 2 CPUs @ 2.0 GHz Pentium 4 or equivalent *Memory 2 GigabytesHard Disk Space Based on customer dataCD ROM Yes* The recommended minimum number of CPUs and minimum memory values are dependent on your data. One of the following operating systems Supported versionsWindows 2003, XP sp2HP-UX 11 11.11.0509. Only for WebSphere 6, you must apply the linker patch PHSS_33035.Solaris 9 5.9 Generic_118558-19 sub4u SPARCSolaris 10 5.10 Generic_118833-03Linux SUSE 9 SP3, Red Hat Enterprise Linux AS 32 or 64-bit

    Siperian Hub XU Rollup Patch A Release Notes

    11 of 22

  • v4.3 (2.6.9-34.0.1.ELsmp)AIX 5.3 (with the IBM JDK v 1.4.2 and Oracle 9205)

    Patch 5300-03 One of the following application servers Supported versionsJBoss 4.0.3 SP1

    NOTE: JBoss is not supported on AIX platforms.WebLogic 8.1.6 (Enterprise Edition required), 9.2WebSphere 5.1.1.6, 6.1.0.3 Software Supported versionsOracle 9i (9.2.0.8), 10.2.0.3 - with database server and network client and SQL*Plus

    (same release). For Oracle 9i, 10.2.0.x use the 10.2.0.2 driver.Note: Oracle JDBC driver 10.2.0.3 is incompatible with IBM WebSphere JRE 1.4.2 onWindows. You need to use Oracle JDBC driver 10.2.0.2 instead when using an Oracle10.2.0.3 database with WebSphere.

    JDK 1.4.2_13 or 1.5 The following Cleanse Engines are supportedTrillium Director 7.7b**Note: In XU Patch B, we also support Trillium 7.8 and Group 1 CDQ V5. We will not support Group 1 V2.7 Business Objects DQXI 11.6(FirstLogic)IBM QualityStage 8.01Platon The Oracle client must be installed on every machine on which you wish to use Hub Cleanse Match. This Oracle client requiresOracle Java utilities, so you must install these utilities on each machine on which you have Hub Cleanse Match as well. The OracleClient and Oracle Java utilities are part of your Oracle distribution. You must also install SQL*Loader on each machine on which you use Hub Cleanse Match Server. SQL*Loader is part of your Oracledistribution. On UNIX systems, the user that starts the application server for the system that hosts the Cleanse Match Server musthave Oracle in the path and Oracle home, base and SID defined in the environment variables. Note: For large datasets, JVM 1.5 is recommended as it has superior memory management features. Using the Platon Cleanse EngineImportant: Configure Platon to use CERTIFY mode, and not CORRECT_ONLY or CORRECT_CERTIFY.To learn more, see your Platon documentation.

    Hub ConsoleHardware RecommendedProcessor (Intel) 1 CPU @ 2.0 GHz Pentium 4 or equivalentMemory 1 GigabyteHard Disk Space 500MBCD ROM No Software SupportedWindows 2003, XP sp2Sun JDK with Java WebStart 1.4.2_13 and 1.5

    (WebSphere 5.1.1.6 requires IBM JDK 1.4.2 for Client and Server; WebSphere 6.1.0.3 requires IBM JDK 1.5 for Server,

    JDK 1.4.2 or 1.5 for Client)Back to top

    Troubleshooting TipsAdmin Console

    Use the following two properties in cmxserver.properties to configure initial heap size and maximum heap size for JavaWeb Start with JBOSS.

    jnlp.initial-heap-size=128mjnlp.max-heap-size=192m

    Siperian Hub XU Rollup Patch A Release Notes

    12 of 22

  • Auto merge Process Records flagged for automerge won't get merged if the match pair is invalid when match revalidate is enabled. If you execute an auto match and merge job, it completes successfully with 1 job shown in the status. However, if you

    stop and restart the application server and return to the batch viewer, you see a second job with a warning a fewseconds later. The second job is Siperian Hubs way of ensuring that either the base object is empty or there are nomore records to match.

    Batch/Tuning

    ANALYZE_BATCH_JOB_THRESHOLD: This value is defined in the C_REPOS_TABLE, the default is 10000. If the number of records exceeds this threshold,the load process will do the analysis.

    PARALLEL_BATCH_JOB_THRESHOLD:

    This value is defined in C_REPOS_TABLE, the default is 1000. The load process enables the parallel degree option ifthe volume is greater than this threshold.

    CASE_INDICATOR:

    Case_indicator is a field on C_REPOS_TABLE. The field only applies to entries that correspond to Packages and isused for Case insensitive searches in GetList and SearchQuery APIs. See API section below for more details.

    Cleansing

    When you refresh your external cleanse, Siperian Hub doesn't take the values from cmxcleanse.properties. So if youchange something in the properties file and refresh, the change you made is not reflected in the product. To fix this, runpostinstallsetup.

    Do not use DeliveryAddressLines and Building input fields in the same mapping. DeliveryAddressLines are for "semi-structured" data and would include the Building data.

    Database Configuration

    Creation of data sources on JBoss 4.0.3 SP1 takes a long time. This JBoss problem was introduced in the 3.2.4release of JBoss

    An exception may appear in the log files when registering the data sources on JBoss. This is caused by thedeficiencies in the deployment architecture on JBoss and results from the server trying to deploy twice. The system isdealing with this situation gracefully.

    The versions of MRM and your ORS must always match. A mismatch, such as an XU version of MRM with an XTversion ORS will cause exceptions.

    Documentation

    The Sample Tutorial Guide, mentioned in the various guides, is not included in this release. Hierarchy Manager

    When converting a base object to a relationship base object, HM does not use the existing column names, nor doesthe interface give an option to select columns. For the relationship base object, HM creates the default column names.HM has specific requirements for column names. This means that mappings may have to be reconfigured.

    Install

    When the cleanse server is not installed on the same application server as Hub, the datasources must be manuallycreated on the Cleanse server to take advantage of connection pooling and improved performance

    By default, HP-UX does not enable SHLIB_PATH lookups for its executables and shared libraries. If shared librariesare not found, even though you have them in your SHLIB_PATH environment variable, enable SHLIB_PATH. Toenable SHLIB_PATH for HPUX you must flip the bit that tells it to use the SHLIB_PATH by doing the following:

    chatr +s enable filenamewhere filename is the name of the executable or shared library

    Prior to installing Siperian Hub, install Oracle with all Java options selected. Cleanse path will fail to load, if it exceeds 256 character or libraries. Uninstallation of the Hub Server for WebSphere and JBoss users: When performing an uninstallation or reinstallation

    of the Hub Server, check for the siperian-mrm-xs.ear ,siperian-mrm-362-was.ear, siperian-mrm-361-was.ear, or thesiperian-mrm-360-was.ear file and delete it if it was not deleted from the WebSphere repository or JBoss deploymentfolder. You can do this through the WebSphere administration console. Also, check the subfolders of the installedAppdirectory for the physical .ear file and delete it if it exists. When re-installing, if the application is installed in WebSphereor if the physical .ear file exists, the Hub Server installer will not set up and deploy the new .ear file in the WebSphereapplication server. You must then manually install and deploy the application. See the Install Guide for moreinformation.

    When starting the WebSphere application server on an HP-UX system, you may see a java.lang.OutOfMemoryErrorwhile deploying or starting up the application server after installation. To fix this, edit MaxPermSize in thesystemlaunch.properties file. This file is in the wasjmx.jar in the WebSphere lib directory. Set the MaxPermSize for

    Siperian Hub XU Rollup Patch A Release Notes

    13 of 22

  • permanent generation space for the JVM to 128 Mbytes. The default value is 64 Mbytes and this may be insufficient foryour system.

    If you are planning to run both the Hub Server and the Hub Cleanse within the same JBoss application server, youneed to follow the instructions in the installation guide to modify the JBoss default classloader.

    JRockit JVM is not supported. Use Suns JDK when installing Siperian Hub on WebLogic. If you install the supporting software (application servers, database, cleanse engines) with the default port settings,

    you might find port conflicts on the following ports:o 1099 (Trillium/JBoss) o 8080 (Oracle Webservices/JBoss).

    To fix this, you must change the conflicting ports. To change the Trillium port:1. Change the Trillium port in the tril7v6\rmi\ports.txt file. Using port number 1069 works without conflicts.2. Run TrilServer.bat

    To change the JBoss port:1. Change the JBoss port in the C:\jboss-4.0.3sp1\server\default\deploy\jbossweb-tomcatxx.sar\server.xml

    Change the port from 8080 to 8888 as follows:

    2. If you find this problem after you have done the installation, you must also change the port number in the mrmserver properties file. For instructions on how to do this, see the Troubleshooting section of the Siperian HubInstallation Guide for your platform.

    3. Run postinstallseup.

    Installing Siperian Hub on WebSphere and JBoss on the same machine causes a communication exception. Toaddress this, in your JRE/lib folder, the IBM Pluggable application client installs a file named "orb.properties". This file isrequired for communication with the IBM WebSphere Application Server. However, it causes an exception whencommunicating with other application servers, such as JBoss. The workaround is to delete or rename the"orb.properties" file. This allows communication with JBoss. This file must be restored if communication withWebSphere Application Server is required. Hub Console fails when connecting to different WebSphere application servers. To address this, in the WebSphereadministration console, change the host property under Servers>ApplicationServers>your_server_name>EndPoints>BOOTSTRAP_ADDRESS to the fully qualified domain name of your application server (E.g.:server1.siperian.com). Save and restart the application server. There is a known Oracle issue when executing setup_ors script on 9.2.0.6 on Solaris. When running loadjava, you mayget a JNI panic error. To address this, use the thin option in the setup/setup_ors script. For example: loadjava -thin-user scott/tiger@myserver:1526:DEMO4536 pass.jar The installer fails to add path environment variable if it exceeds 256 characters. When you enter the unlock code for Platon in the install UI, an incorrect unlock code is saved into properties afterinstallation. If your unlock code has the $ or / characters, it might not get interpreted correctly by the UI.To address this, after you complete the install process, edit the cmxcleanse.properties file so it correctly reflects yourunlock code.

    JBoss

    JBoss fails to start if the cmx_system user is named anything other then the default. This only applies to the JBossapplication server. If this happens, change the siperian-mrm-ds.xml file manually.

    Landing tables/Stage process

    INTER_SYSTEM_TIME_DELTA_SEC:If records in the landing table contain future times, even just a few seconds into the future, the stage process would setthat LAST_UPDATE_DATE in the staging table to 1/1/3000. And during the load process these records would not beinserted or update base object tables. The INTER_SYSTEM_TIME_DELTA_SEC can be used in such cases where there are discrepancies between theAppserver machine and the database machine. This column is available on the C_REPOS_DB_RELEASE table. Thedelta value specified in this column is in seconds and is used to calculate the last update date (LUD).

    Siperian Hub XU Rollup Patch A Release Notes

    14 of 22

  • This setting only applies to the Stage process. The load does not use this setting.

    Load Process

    The threshold for parallel degree for loading is configurable. In the C_REPOS_TABLE, there is a column whosevalue you can set to configure parallel degree for loading:PARALLEL_BATCH_JOB_THRESHOLD the default value is 10000.

    Logging

    If you are using UNIX, debugging is turned on, and cmx_debug.log exceeds the UNIX file size limits, debug prints willnot appear in the log once you hit the 2G limit on UNIX.Workaround: Move or delete cmx_debug.log.

    Match Summary logging has been moved to the INFO level, so this information is available when logging is not set to the

    DEBUG level. Segment match values are case-sensitive. When using segment match on fuzzy and exact base objects, the values

    the user sets are case-sensitive when executing the match process. This is different from the behavior in XS. Property to enable large rowid_objects during Match:

    Message Queues

    Support for direct IBM WebSphere queues has been discontinued in Siperian Hub XT. If you have already configuredIBM WebSphere queue in a previous Siperian Hub release, using IBM WebSphere, you must drop and create thequeue again with IBM WebSphere MQ through their JMS interfaces. To configure the queue with MRM see UsingMessage Queues in the Siperian Hub Administrators Guide.

    Migration

    When you are migrating from one environment to another, if you do not have sufficient privileges, you may see thefollowing error on staging:Current status: Couldn't start procedure execution: SIP-24200: Error: start_cleanse failed with ret_code:-20904,errmsg:java.security.AccessControlException: the Permission (java.net.SocketPermission host resolve) has not beengranted to YOUR_ORS. The PL/SQL to grant this is dbms_java.grant_permission( 'YOUR_ORS','SYS:java.net.SocketPermission', 'host', 'resolve' ) [http://host:8088/cleanse/] Runtime Exception cmxcl.start_cleanse You can avoid this by either giving the user running the stage job system privileges, or by running the following as asys or DBA user:execute dbms_java.grant_permission(upper('YOUR_ORS'),'SYS:java.net.SocketPermission','*','connect,resolve');.Here, YOUR_ORS is the name of your ORS.

    Merge Manager, Data Manager

    In order for a non-admin user to see packages in the Data Manager, they must have permission to use the toolassigned by the admin. Non-admin users must also be granted the roles for the put packages (both get and put).

    Oracle

    Merge can fail unexpectedly during Match processing if the Oracle TNS Listener refuses connections. This is aknown Oracle issue. To address this, increase the processes on Oracle.

    The index on STRP becomes unusable after first load. This may happen due to an Oracle bug when upgrading from9.2.0.1 to 9.2.05. Apply the catcio.sql

    Incorrect Driver version: If you see pervasive errors in various places, Oracle JDBC Driver may need to beupdated to v. 10.2.0.2.0.

    Parallel degree on Oracle 9i: The recommendation is for Customers to set Parallel=1 if they're on 9i the followingerror may be thrown if Parallel > 1ORA-12828: Can't start parallel transaction at a remote site

    Other

    You can receive an out of memory exception when you import a very large number of match strings. The number ofimported strings that will trigger this error depends on your data and the memory configuration of your machine(s).

    Platon Address Doctor

    It is recommended that you Use Certify mode only. This property is set in/resources/cmxcleanse.properties (on UNIX) and \resources\cmxcleanse.properties (on Windows)

    Schema Manager

    Siperian Hub XU Rollup Patch A Release Notes

    15 of 22

  • When importing a column of data type NCHAR of NVARCHAR, its length will be the byte size rather than the charsize. So, if you import a NVARCHAR column whose char length is 50, it becomes 100 when imported. To address this,import the column and then change the length to the char size.

    If you have two base objects (parent and child) with the child having a foreign key Rowid_Parent; when you create

    a match column on Rowid_Parent on the child table with the exact same physical name that is stored in the base object(Rowid_Parent) and attempt to use any child match column in an inter-table match in the parent table, it fails to save. An exception displays Unable to commit matching table C_CHILD_STRP because there is already a match columnnamed Rowid_Parent. To address this, rename the Physical Name of the foreign key match column used in the childmatch columns.

    Unicode

    If you are using the data type varchar or char for double byte strings, you might receive a validation error duringPUT. To address this, use the data type nvarchar or nchar instead.

    WebSphere

    If you are using the WebSphere application server, you may encounter a memory constraint when you deploy yourapplication. To fix this,

    Windows1. Open the WebSphere ejbdeploy.bat file for editing. By default, this file is in: C:\Program Files\WebSphere

    \AppServer\deploytool\itp\ejbdeploy.bat2. Edit the following line:

    "%JAVA_HOME%\bin\java" -Dwebsphere.lib.dir="%WAS_HOME%\lib" -Dws.ext.dirs="%WAS_EXT_DIRS%" -cp"%ejbd_cp%" -Xms256M -Xmx512M com.ibm.etools.ejbdeploy.EJBDeploy %*

    3. Add the arguments in bold (-Xms256M -Xmx512M).4. Save and close the ejbdeploy.bat file. UNIX1. Open the WebSphere ejbdeploy.bat file for editing. By default, this file is in: /opt/WebSphere/AppServer

    /deploytool/itp/ejbdeploy.sh2. Edit the following line:

    $JAVA_CMD -Xms256m -Xmx512m -Dws.ext.dirs=$WAS_EXT_DIRS -Dwebsphere.lib.dir=$WAS_HOME/lib -cp $ejbd_cp $EJBDEPLOY_JVM_OPTIONS com.ibm.etools.ejbdeploy.EJBDeploy "$@"

    3. Add the arguments in bold (-Xms256M -Xmx512M).4. Save and close the ejbdeploy.bat file.

    Postinstallsetup fails on WebSphere:

    This usually happens if the Siperian queues were not configured properly before the Hub was installed.

    When the listener service is not started on WebSphere 6:The Siperian Hub posts the messages successfully, but the MDBs that are supposed the pick the messages up havenot been started, so we cannot process the calls

    TestConnection fails on WebSphere:WebSphere needs to be restarted after creating datasource for ORS from console

    WebLogic

    Increase Stuck Thread Max Time in WL if long running batch processesThis property is usually under the AdminServer tuning (go through the WL console)For example: Home > Summary of Environment > Summary of Servers > AdminServer

    If this value is not increased, then WL does self-tuning when the time exceeds this limit. Usually, sqllloader and otherbatch processes (tokenize, etc.) take a while and WL might close connections in between before the processcompletes. Interesting errors happen during match if the connection is closed before the process completes.

    SIF

    When using the GETLIST verb and passing in a DATE datatype, the default format for Oracle is "DD/MM/YYYYHH24:MI:SS"

    On WebLogic, using cascade unmerge through SIF API can cause timeout errors. Fix this by setting theServices/JTA/Timeout Seconds in the WebLogic app server.

    As of XT SP1, null strings are handled in a consistent way between the PUT and GET verbs. Both PUT and GETnow represent an empty string as a NULL value.

    Data is not merged if the PUT and MERGE verb calls are in the same transaction. To address this, when executingthe merge verb in a transaction, you must perform a commit before the merge verb is called if the data prior to the

    Siperian Hub XU Rollup Patch A Release Notes

    16 of 22

  • merge is required. The merge verb will not be able to see the changes from the PUT if done in the same transaction. Execute the merge separately.

    When using the SIF API with WebLogic, you may see the following exception thrown: OracleXAResourceXAException XAER_PROTO. This may be caused by connections being closed before a transaction is complete. Toaddress this, use the recommended Oracle driver.

    The MATCH verb will fail if you set security ON. To address this, turn off security. SIF ORS API Generation:

    SIF ORS API generation can fail if you have bad metadata. For example, duplicate entries inC_REPOS_CL_FUNCTION because of an old schema manager bug where we didnt drop the entries when a mappingwas dropped. The easiest way to clean this up would be to update the function name from the UI, see which entry itupdated in C_REPOS_CL_FUNCTION and remove the dupes that are not relevant.

    SIF/Tuning METADATA_CACHING_ENABLED_IND:

    Metadata caching is disabled by default on the CMX_SYSTEM and ORS. It can be enabled for the purpose ofimproving XML/SIF API performance on a production machine.

    To enable caching for the Master or ORS:update c_repos_db_release set metadata_caching_enabled_ind=1;

    To disable caching for the Master or ORS: update c_repos_db_release set metadata_caching_enabled_ind=0;

    After enabling or disabling caching, all MRM servers and MRM-Cleanse servers connected to that Master or ORS mustbe restarted or reinitialized. If MRM metadata caching is enabled for the Master or an ORS, the Console can not beused to make changes to the configuration, so trying to acquire a Write Lock on it will give an error dialog.

    Back to top

    Resolved Issues and Enhancements Resolved Known Issues

    14927: Cannot remove a ruleset from a base object if it is the only ruleset available.14990: The match column isn't shown as a SubType match column as it is when the user enables subtype using the editmatch rule dialog.14991: Removing a ruleset filter does not clear the ruleset filter sql, therefore when the filter is enabled again, it will containthe previously removed SQL13331: Platon cleanse engine initialization fails on AIX.6848: If you define the same column more than once in the display package, you will see an error in the Merge Manager.9611: Installer deploys to any running domain if the one specified is not running. If you specify a domain during installationand that domain is not running when you invoke the postInstallsetup process, postinstall deploys to any domain that isrunning at that time.10827: The application server port is not exposed in the build.properties file for Hub Server. This is exposed for HubCleanse Match server. For Hub Server, the RMI port is exposed. If you change the RMI port, re-deploying the applicationsucceeds, but when you start the console, it fails as the appserver port still has the old value.11965: On Solaris 9, using the WebSphere application server, the upgrade of the cleanse match server can fail at thepostinstallsetup script stage.13320: If you have a child base object which has 2 lookup columns, one from Parent 1, second from Parent 2, when you runthe stage process, an index is created on the first lookup column but not on the second. This may have a negative effect onperformance.13069: With two instances of Hub Console running, you cannot concurrently run a manual merge while a manual unmergeis running on the same data.10078: Unlike the Data Manager, the API does not use the Allow Null Foreign Key flag. The PUT call inserts or updates therow, ignoring the value of that flag.12533: If you set the encrypted flag to true for a password, you must encrypt the password yourself before using the stringin the XML you are passing to Siperian Hub. There is no facility in Siperian Hub to encrypt this password for you.13389: Search queries fail if the generated query is too long.13407: You can receive a null pointer exception in the following circumstance: Enable trust and execute a GET SIP API callto get Package, XREF and lineage information.12612: If you are using Oracle 9i, unmerge can fail but appear to succeed in the following situation:1. Load base object with a non-nullable column.2. Execute a Put XREF for an existing rowid_object.3. Go to the data manager and unmerge the new record.The data manager says that it was done successfully when in fact the unmerge was rolled back because of the NULL.

    Enhancements:

    Siperian Hub XU Rollup Patch A Release Notes

    17 of 22

  • (4 digit numbers represent support case IDs, 5 digit numbers internal requests) 3142, 3144: Process (job) flow capability added in the form of Batch Groups

    2734: Unmerge is now transactional2769: PL/SQL Error Message now transferred to the Error Message Box on MRM Console2883: Rule information now returned for exact match3030: Support for VARCHAR2(4000) from console UI now available

    3148: Open batch process nodes now do not collapse when you select another process3241: BO LUD now populated on insert of new record, not just on update3326: All "Analyze table" statements have been converted to dbms_stats.gather_table_stats for data3057: Unmerge merged BOs with XREFs now migrates old XREFs back to the original BO3238: Unmerge child on parent unmerge now works for children with multiple parents3394: Metadata Validation tool made available in the console3577: Now able to clone an existing landing to staging mapping for another source3393: Able to choose which Rule Set to use when executing Match Batch jobs3497: Metadata validation now identifies invalid constraints e.g. columns indexed twic3581: Match and Link support configurable at BO level3382: Fetch Many Hops Limit in HM is now customizable3428: Non-system errors can now be raised from user exits from HM screens3880: Ruleset can now be removed even if it's the only ruleset available3140: Lookup column is now nullable whenever allow null foreign keys is selected4240: SIF developer now has more control over transactions4367: Performance improved for multimerge14763: Allow the ability to select a specific rule set during a Match Batch Job.

    15033: Ability to have secure JNDI nodes10500: Relationship viewer now shows show pkey_src_object and source system10528: The upd by, created by, upd date cols are now displayed in the history viewer13863: Metadata validation will check if the query sql is out of synch with the C_REPOS_QUERY_COL entries

    Back to top

    Known IssuesNote that all numbers listed below are Siperians internal Problem Incident Reporting system (PIR) tracking numbers. Please refer tothis number when making enquiries to Siperian Support.

    Batch Group Management

    15713: A warning from a job within a Batch Group will cause the Group Job to return a status of Completed With Errors. This error message returned by the Batch Group should be investigated by viewing all statuses within the group and can beignored.15949: When converting a Link-Style to a Merge-Style Base Object, the Auto-Link job will not be automatically removedfrom the Batch Group. This will need to be done manually.

    Cleanse Process

    7512: Siperian Hub does not connect to next available cleanse server if the cleanse_url is invalid.7592: The cleanse daemon does not go offline automatically once the sqlloader command fails so that consecutive requestsdo not go to this server.9634: Under certain circumstances, you may have to recreate mappings created using Trillium. If you create a map using aTrillium component, then switch to a cleanse server that doesn't have that Trillium project available, when you go back tousing the Trillium library and the map you were using, the function is removed. You may also see functions lost if you restartthe application server for Hub Cleanse Match and Trillium fails to initialize for some reason. After you address the issue thatcaused Trillium initialization failure and restart the application server, you must recreate the mapping.10835: The start_cleanse function runs asynchronously if in_run_synch is not TRUE. This value is case-sensitive.12785: When testing Trillium Director on WebLogic 9/WebSphere there is an issue with the Trillium Directory shared librarythat may cause a core dump. This issue has been raised with Trillium Support.13227: The cleanse function subtract floats (floating point values) gives inaccurate results.13255, 14562, 17009 : FirstLogic cleanse engine requires WebSphere 6.0.2.3 and above. The DnB cleanse engine doesnot work with WebSphere App Server since DnB doesnt support IBM JDK.15069: Acxiom cleanse engine is not compatible with WebSphere 5.x or 6.x16368: Platon may fail, causing a core dump on the application server. This will cause the staging job to fail the issue hasbeen raised with the vendor; waiting for a fix/release from the vendor.

    Data Groups This tool has been deprecated.

    Data Manager, Merge Manager

    Siperian Hub XU Rollup Patch A Release Notes

    18 of 22

  • 9697: Incorrect results when you search on lookups.1) You can choose different search types: is exactly, begins with, ends with, contains, does not contain, is like, is not like.This is chosen from a drop-down - similar to the Search Wizard.2) A page of 100 results is returned. If there are more than 100 results the user can page through the entire result set.3) When the number of lookup values is below a set limit they are displayed in a drop-down list - when there are more thanthe limit the lookup search is used. This limit can now be configured at run-time by setting the sip.lookup.dropdown.limitparameter in the cmxserver.properties file. The default for this value is 100.8526: A manual Merge/Unmerge process does not terminate in a clustered environment when one of the managed serversis down.Workaround: Mark the process as incomplete and repeat the merge/unmerge.13211: If you are running on a clustered WebLogic 9 application server, you may receive an unmarshalling exception whenscrolling through records in the data manager.Workaround: Stop the process running the Hub Console and restart it. Alternatively, wait and the problem will resolve itself.13358: A user who does not have Siperian administrator rights can set a batch job status to incomplete.

    Delta Detection 10970: Online previous landing (OPL) tables are used for delta detection for CLEANSE_PUT operations. The size of thesetables is not managed by Siperian Hub. In order to prevent the tables from becoming excessively large, you must managetheir size manually.

    Environment Staging9764: The environment staging process migrates the MQ rule metadata. This can cause a problem if the target environmentis configured with a different CMX_SYSTEM and a different MQ Server. This situation will cause a null pointer exceptionwhen trying to load the MQ Rules in the Schema Manager.

    FirstLogic13099: If you are using HP-UX with the FirstLogic adapter, you will receive an error. This is an issue with the FirstLogicinstallation. The HP shared object links in the incorrect runtime environment files. Workaround: You must install the 'Runtime Environment for International Address Engine' to workaround this issue.

    HM

    10092: After adding relationships, the relationship count may not be updated. This is benign.10639: It is not possible to sort on entity grids.11808: The view related entities feature can give inconsistent results if you have sandboxes. This issue applies to the viewfeature; it does not reflect any inconsistency in the underlying data.12902: You must include all HM required columns in the staging table, or the load process will fail.

    Install

    8529: If database changes are applied after the ear is deployed, the application server must be restarted to refresh thecache.8753: If you have an invalid debug_file_path and debug is enabled, it will fail to apply the2004102202_ldc_ConInd_DirtyInd_change.sql9199: Null pointer exception thrown intermittently when launching the console. This error is benign and has no effect on theoperation of the console.9673: If you choose to not run postinstall setup, the Installation logs indicate the installation failed. This is not the case. Thiserror is benign.12031: The postinstall process fails on WebLogic 9 under AIX.Workaround: After installing Hub Server and Hub Cleanse Match, you must manually deploy the Hub Server and HubCleanse Match .ear files within WebLogic. The .ear file for Hub server is siperian-mrm.ear and is located in the rootdirectory for Hub Server. The .ear file for Hub Cleanse Match is siperian-mrm-cleanse.ear and is located in the rootdirectory for Hub Cleanse Match.12276: When you create a new ors, Siperian Hub uses the CMX_DATA and CMX_INDX tablespaces.12453: The installer adds two extraneous files to the Siperian Hub for Oracle installation. These files are create_ors.bat andcreate_system.bat and they apply to the DB2 version of Siperian Hub. This is completely benign.13002: If you are using the WebSphere application server, you must stop and restart the application server after creating adatasource for an ORS from the Hub Console. Otherwise, the newly-created ORS datasource won't work.15188: If you provide the incorrect path for JBoss during a silent install, the postinstallsetup script will say the install wassuccessful even though the installation failed with the wrong JBoss path. Note: There is no validation in silent install.16198: During a first installation on WebSphere 6.1, the postinstallsetup fails for both auto and manual options.Workaround: Restart the application server, go to the AppServer Console and verify the siperian.mrm.ear exists. The file willthen show active status.16354: During a first installation on WebSphere 5.1.1.6 on Solaris, the postinstallsetup script may fail.Workaround: Restart the application server and run postinstallsetup manually.16390: The XU versions of the Siperian Hub documentation suite are not installed into the Siperian\Hub\Server\Documentation directory. For those upgrading to XU, this folder will still contain the previous version of Siperian Hub

    Siperian Hub XU Rollup Patch A Release Notes

    19 of 22

  • documents. Users can find all Siperian documentation in the external documentation folder found on the CD or FTP site.Please contact Siperian Support for more information.

    Load Process

    9933: Load process aborts rather than rejecting records with a date in the future. The assumption is that once records havebeen staged, no invalid dates should exist in the staging table.13337: Load can fail with a deadlock error. This is an Oracle bug that may happen on certain platforms (not all).Workaround: Set Parallel=1 on the base object.

    Match Process

    Match Validate has been deprecated.10298: The consolidation indicator is incorrect after a match executed with no rules set. This error is benign. The recordswith the inaccurate consolidation indicator are picked up in subsequent match jobs.10928: The ROWID_USER columns are not populated in the match table.11412: If you have records on hold (consolidation indicator = 9), as from a Match Analyze operation, refining your matchrules should trigger the Reset Match, resetting the records that are on hold. This expected behavior does not happen.Workaround: You must reset the consolidation indicator to 4.12450: The generate match tokens process can fail with a stack overflow error when the match key (set in the matchcolumns UI) is set to a column of a child base object and that child base object contains approximately 500 or more recordsthat belong to a single parent record.12829: If you are using a clustered WebLogic configuration, and you bring a managed server down while a job is inprogress, the job does not terminate. For stage and match jobs, an error code is written to the database log. For loads andmerges, an error is sent to the console.Workaround: Mark the process as incomplete and redo the match 14065: If your tokenization table fails due to lack of tablespace, you will not be able to recreate it. You can manually solverecreate the STRPT table with this command alter index SVR1_7UW_TMP rename to SVR1_7UW alter TABLE c_address_strp rename constraint SVR1_7UW_TMP to SVR1_7UW; The index names are unique so they wont be the same every time.

    Merge

    12555: If you are using the WebLogic application server with Oracle 9i, have the Parallel Degree set to 4, and run anautomerge with match validate turned on, the following Oracle error will be thrown: ORA-12828: Can't start paralleltransaction at a remote siteWorkaround: Set parallel degree to 1.13043: In the Merge Manager, there can be an issue with the status of an unmerged in the following situation: if you have aset of unmerged records and you click begin merge, all records are returned correctly under Unmerged tab. If you thenhighlight the last row, click the button "change the status for the selected record", and click Put on hold, you may seefollowing message:Index: 9, Size: 9. In the console log, there will be an error: [AWT-EventQueue-0] ERROR.15813: Automerge in batch jobs fails if a base object has more than 200 columns. This is a limitation of automerge.Currently Siperian supports up to 200 user columns in the automerge process.

    Message Queues

    6747: The Schema Manager doesn't allow you to select which columns you want to monitor on an Insert trigger rule as itdoes for an update trigger.13070: If you have message queues configured, operations in the schema manager can take a long time. The reason this isthat for each data change that is associated with a message queue message, the monitoring thread will query the messagepackage.Workaround: Manage the volumes in the message queue tables.16403: The event triggers do not exist for the new Link/Unlink feature functionality.

    MET Validation15411: When you upgrade a schema to XU and then do MET validation, the metadata manager check for views may not bein synch with database views. MET will display a warning. The workaround for this is to re-save the query.15788: During metadata validation on an upgrade from an earlier schema, MET does not validate all triggers properly. If atrigger is invalid, the user cannot create or edit match columns in the schema.15917: When you do metadata validation on a new schema, MET does not detect missing PROXY_ROLE privileges.15918: During metadata validation MET does not detect tablespace related inconsistencies. This might happen if userschange the tablespace after the table is created.16113: When upgrading any pre-XU schema, metadata validation may fail with invalid packages on the first attempt.Workaround is to repeat the validation.

    Migration

    Siperian Hub XU Rollup Patch A Release Notes

    20 of 22

  • 9645: When migrating from MRM 3.6.2 SP1 or Siperian Hub XR to Siperian Hub XS or above, Siperian Hub inserts dummyvalues into the staging src_rowid. This will cause an error when populating the reject table. To correct this situation, repeatthe stage process to generate the correct src_rowid.10392: The migration script does not remove legacy DataFlux pipelines.11593: If youre migrating from XT to XU, you must resave your exact match rules to properly update the views correctly.11621: If you are migrating from XR to XT SP2 and have Leave Match Hit enabled in the pipeline, after migration thecleanse list setting for "Leave Match Hit" is incorrect.

    Oracle16144: In AIX, the database process is not getting released when the message queue message gets drained. Theworkaround is to apply the patchset provided by Oracle. Download the patchset for this issue (Oracle version 10.2.0.3) fromthe Oracle metalink: https://metalink.oracle.com/metalink/plsql/f?p=200:101:4830616257230449809&notification_msg=Invalid%20Login%20Credentials/76A7A626D755801A6A7C8BDC44715B73/

    ORS Users and Groups9749: All ORS groups not listed due to unknown error. If the schema has multiple users in the C_REPOS_USER table thatdo not exist in the CMX_SYSTEM you may see this behavior.12324: For Global or Private Password Policies, the number of grace logins is ignored.

    Other

    9332: Removing the datasource from Siperian Hub does not remove the datasource entry from WebLogic.10910: In a clustered environment, if the application server that received a request to match, stage, or generate tokens fails,the match, stage or generate tokens job does not terminate.Workaround: Terminate the jobs manually.12555: A combination of WebLogic, Oracle 9i and Parallel Degree set to 4 may cause the following error to be thrownduring batch jobs: ORA-12828: Can't start parallel transaction at a remote site. This is due to an Oracle 9i issue that wasresolved in version 10g.Workaround: Set the Parallel Degree to 1.12764: In the metadata, the updated_by field of cmx_ors.c_repos_user is always set to CMX.12767: When you update a user to assign a different ORS to that user, the update may not be reflected in the metadata.13367: Datasources still exist after unregistering the database in WebSphere 6 application server. If no longer required,users can manually remove them via the WebSphere Administrator Console

    Roles

    15560: SAM roles may be converted when upgrading from XT to XU. Verify permissions for roles after upgrade.

    Schema ManagerRelationship tables have been deprecated. Use relationship base objects for this functionality instead.6874: The default value is blank after you enable the Has Default check box. If you add a default value for one column,and then enable the "Has Default" checkbox for the same column, the default value you entered is gone and the check boxis not enabled.8857: If saving the match rules fails because of network outages or connectivity issues, the indexes may get dropped on theSTRP tables. Contact Siperian support.10258: It is possible to create nonsensical match rules, such as a match rule with 1 exact match column with segmentmatch. Siperian Hub will not stop you from doing this.10923: After creating a Staging Table, the "Edit Lookup Column" button is enabled for ALL columns but should only beenabled for those which have a lookup defined.16404: If GBID is enabled and an invalid length value is entered, the message returned by the UI states that a length of 100is required. This is not true, a value of 255 is the expected length.

    Security Access Manager16118: Some changes to the Security Providers setup require a server restart in order to take effect other types ofchanges do not. For example, changes to Login Modules (enable/disable, moving providers up and down) take effectimmediately, while changes User Profile and Authorization providers require a restart.16310: When using SAM Role Report, there may be incorrect information given regarding rights from the parent role. Thisonly affects the report when you have multiple parent roles.

    Siperian Sample15613: To use the Siperian Sample installer and schema you must have a valid license for Hierarchy Manager in the Hub.16237: When manually deploying the sipsample.ear file, the deployment name will be sipsample (${sip.version} build ####release XU. This is a labeling issue and will not affect functionality and usage.15980: Deployment of the Siperian Sample will fail if the Sample application is already deployed (upgrade/update).Workaround: Manually undeploy the Siperian Sample application via the application server Console and rerunpostinstallsetup

    Siperian Hub XU Rollup Patch A Release Notes

    21 of 22

  • Security12296: When enabling Secure IIOP transfer (IIOPS), the ORBTC log file will grow drastically on the desktop where theSiperian Console is launched - this is a known IBM WebSphere issue. Workaround: Add a fake certificate to theSiperianKey JKS file.

    Services Integration Framework (SIF)7113: The GETLIST verb ignores the page size if the page number is 0. The number of rows returned default to theGETLIST default page size (200). If the page number is 1 or greater, it returns the page size requested.10330: When you use the PUT verb, Siperian Hub does not reject rows with an invalid value in the dep_rowid_systemcolumn.10981: When using the MATCH API call on an exact match base object without rule set/rules, the result message indicatessuccess when this is not the case.10995: For an exact match base object, the rule number isn't included in the XML response message.11859: Match verb fails if you have data group security on.Workaround: Turn data group security off.12015: The MATCH API disregards the value of the distinct system setting on the base object.12805: You cannot use setRecordState within a transaction.13013: One or more exact rules on a fuzzy base object dont return rows for FAUTO or FBOTH.13371: The generation of ORS-specific SIF APIs fails if the JDK is not available in the classpath.Workaround: Add the JDK to the classpath.16287: The SAM privileges required to execute searchHMQuery request are HM Profile READ, Package (columns) READ. In this release, removing the READ privileges on specific columns does not work properly i.e. all columns in the Packageare returned regardless of whether the User has column READ access or not. The workaround is to create a newPACKAGE which only includes the required columns.16251: The Multi-merge call cannot rollback changes because a commit takes place after every merge.16377: The PUT API will fail when attempting to execute a PUT XREF with GBID enabled.

    Tool Configuration12767: When you re-assign users to an ORS using the USERS tool, the creator and the updated_by values are notupdated in the ORS.

    Uninstall

    15498: When using the uninstall feature of the Hub Server Installer, the uninstaller may hang while running thepatchinstall/uninstall script. Workaround is to reboot the machine.

    Unmerge

    11378: An unmerge process can fail because of resource contention, for example table locks. If the unmerged objects havechild objects, this unmerge failure can result in inconsistent child objects.15400: Cascade unmerge ALL fails in a grandparent/parent/child scenario, a three-levels+ unmerge of records.16154: If child records are loaded after merging parent records, the lineage data needed to cascade unmerge child recordswill not be populated, which means the cascade unmerge for child records will not function.Contact Support for a user exitworkaround.

    Unicode

    10427: A base object column of type nvarchar or nchar is not available to be selected as a match column. Only varcharcolumns are available to be match columns.10434: If you have the regional font set to something other than English, when youre using the Mapping tool, the text on thewarning dialogs