Duplicate computer system problems

This information covers the problems that occur when duplicate computer systems are created but never resolved by TADDM during discovery.

General problems

During discovery, duplicate computer systems are created, and this behavior is expected. Duplicate systems are typically created when a sensor creates a configuration item (CI) in the database but does not have enough information to uniquely identify it. For example, assume during a discovery, both the Stack Scan sensor and the Ip device sensor learn that an IP device exists. However, the sensors cannot uniquely identify the device. For example, if a device has two network interfaces, the sensors cannot identify if it is one device with two interfaces or two devices with one interface. Another example is a VMware sensor that has enough information to uniquely identify the computer. However, the information is different from information reported by another sensor for the same computer.

If a credentialed discovery is performed after running a credential-less discovery, objects created might not be reconciled. Check the MSS tab of any duplicate objects to confirm if a custom application sensor and a specific application sensor discovered the object.

Sensors provide the attribute data that they can collect for a CI. However, if there is insufficient information to identify that the naming attributes match or overlap between different copies of a CI, then it is not possible to determine that the CIs are the same object. The result is duplicate objects.

In most cases, logic in the TADDM server can resolve and eliminate duplicate computer systems while they are being stored. In some circumstances, for example when Level 1 and Level 2 discovery sensors are enabled in the same profile and a computer is discovered for the first time, duplicates can be stored temporarily. These duplicates are resolved by background reconciliation after a few minutes. When duplicate computers remain beyond several minutes, TADDM has insufficient information to automatically resolve these duplicates. You must merge these duplicates in the Data Management Portal. For more information, see Discovery tasks.

If the duplicates have matching signatures, send the following information to IBM® Support:

The API query results. To obtain these results, run the following command:
```
api.sh -u user_id -p password find "Select * from ComputerSystem"
```
The topology manager and discovery logs, with the logging level set to DEBUG
The results files ($COLLATION_HOME/var/dwitem/result/* files) from the discovery in which the duplicate computer systems were created

Merging CIs within an access collection based on a custom query might cause inconsistencies

Problem

Merging CIs within an access collection based on a custom query might cause duplicate instances in the COLLECTIONJDO_MEMBERS_X database table. These duplicates can lead to inconsistencies and errors in the UI. If a merged CI in an access collection is also a member of another collection, it might be removed from the other collection.

Solution

Use an SQL statement to remove all duplicates from the database. The following statement can be run, on one line, on DB2® or Oracle databases:

DELETE from COLLECTIONJDO_MEMBERS_X where (PK__JDOIDX, MEMBERS_X, JDOORDERX)
 in (SELECT PK__JDOIDX, MEMBERS_X, JDOORDERX from COLLECTIONJDO_MEMBERS_X, 
	(SELECT PK__JDOIDX as JDOIDX, MEMBERS_X as MEMBERS, MIN(JDOORDERX) 
	as MINORDER FROM COLLECTIONJDO_MEMBERS_X GROUP BY PK__JDOIDX, 
	MEMBERS_X HAVING COUNT(MEMBERS_X) > 1) 
TOLEAVE where COLLECTIONJDO_MEMBERS_X.PK__JDOIDX = TOLEAVE.JDOIDX and 
COLLECTIONJDO_MEMBERS_X.MEMBERS_X = TOLEAVE.MEMBERS 
and COLLECTIONJDO_MEMBERS_X.JDOORDERX > TOLEAVE.MINORDER)