APAR status
Closed as program error.
Error description
During Oracle RAC Discovery, the oracle instances are not linked to the Oracle RAC object or linked instances gets deleted after a background agent run. For this the fix involves logic changes and setting of following two properties in collation.properties file on Discovery Servers com.collation. oracle.sensor.rac.UseHostnameOfCrsStatForMerge=true com.collatio n.oracle.sensor.rac.AccessOtherScanNodes=true During a discovery run of a single node of Oracle RAC, Oracle Instances are discovered by using data of crsstat command and by using data of "ps -ef|grep pmon" command. The instances from these two command repeat for some data. The data from crsstat command only contains short hostname , which is usually not the Fully Qualified Domain Name (FQDN) of the host. So the first property id used to merge the two instance based on the same SID+HOME data and then using the hostname of crs_stat in the merged data. This hostname is needed to be set because the same OracleInstance discovered from other RAC node discovery of same RAC will be merged using this hostname which is part of naming rule of OracleInstance The second property of above is needed because, when we scan an Oracle RAC node, we also try creating instances of other nodes if we are able to find the oracle home of those instances. The "oracle home" is tried using multiple approaches, first using JDBC connection , then reading configuration files. In many cases the lookup of oracle_home of instances belonging to the other nodes are being tried (in case of lookup from configuration files ) only from the node for which the sensor is running. But we need to check the configuration files on the other nodes, so we have tried creating an additional OS connection to other RAC nodes to fetch configuration file. Although currently discovery is run for all nodes , but they only replace data of one another in terms of linking of OracleRAC object with the OracleInstances . So we are trying that each individual scan of a node fetches more data so that even after replace more OracleInstances are linked to Oracle RAC.
Local fix
NA
Problem summary
During Oracle RAC Discovery, the oracle instances are not linked to the Oracle RAC object or linked instances gets deleted after a background agent run. For this the fix involves logic changes and setting of following two properties in collation.properties file on Discovery Servers com.collation.oracle.sensor.rac.UseHostnameOfCrsStatForMerge= true com.collation.oracle.sensor.rac.AccessOtherScanNodes=true During a discovery run of a single node of Oracle RAC, Oracle. Instances are discovered by using data of crsstat command andby u sing data of "ps -ef|grep pmon" command. The instances from these two command repeat for some data. The data from crsstat command only contains short hostname , which is usually not the Fully Qualified Domain Name (FQDN) of the host. So the first property id used to merge the two instance based on the same SID+HOME data and then using the hostname of crs_stat in the merged data. This hostname is needed to be set because the same OracleInstance discovered from other RAC node discovery of same RAC will be merged using this hostname which is part of naming rule of OracleInstance. The second property of above is needed because, when we scan an Oracle RAC node, we also try creating instances of other nodes if we are able to find the oracle home of those instances.The "oracle home" is tried using multiple approaches, first using JDBC connection , then reading configuration files. In manyc ases the lookup of oracle_home of instances belonging to theo ther nodes are being tried (in case of lookup fromc onfiguration files ) only from the node for which the sensor is running. But we need to check the configuration files on theo ther nodes, so we have tried creating an additional OSc onnection to other RAC nodes to fetch configuration file.Although currently discovery is run for all nodes , but they only replace data of one another in terms of linking of OracleRAC object with the OracleInstances. So we are trying that each individual scan of a node fetches more data so that even after replace more OracleInstances are linked to OracleRAC.
Problem conclusion
The fix for APAR is contained in the following maintenance packages: | Fix Pack | 7.3.0-TIV-ITADDM-FP0008
Temporary fix
Comments
APAR Information
APAR number
IJ26967
Reported component name
APP DEPENDENCY
Reported component ID
5724N5500
Reported release
730
Status
CLOSED PER
PE
NoPE
HIPER
NoHIPER
Special Attention
NoSpecatt / Xsystem
Submitted date
2020-08-12
Closed date
2020-08-27
Last modified date
2020-08-27
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
Modules/Macros
999
Fix information
Fixed component name
APP DEPENDENCY
Fixed component ID
5724N5500
Applicable component levels
[{"Business Unit":{"code":"BU053","label":"Cloud & Data Platform"},"Product":{"code":"SSPLFC","label":"Tivoli Application Dependency Discovery Manager"},"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"730","Line of Business":{"code":"LOB45","label":"Automation"}}]
Document Information
Modified date:
28 August 2020