Service Graph Connector for Microsoft SCCM - Partial payloads and IRE

[Palani Chockali]

Link to understand IRE partial payloads:

https://docs.servicenow.com/bundle/rome-servicenow-platform/page/product/configuration-management/concept/ire.html#d1561248e246

IRE will create partial payloads for disk, processor or software or other entities in the partial payloads table (cmdb_ire_partial_payloads) if it has reference to a computer which does not exist yet in CMDB for the discovery source 'SG-SCCM'. This is acceptable behavior. However, there is a known issue with IRE that it could cause performance degradation for large partial payloads and merging. This is being addressed by the IRE team and not in place yet.

In some of the customer cases, the reason partial payloads are being created is because a full pull of computer was not performed. This could be either during the migration from SCCM 2016 or even in a straight up installation.  If the 'Use last run datetime' was checked during a computer pull for a first time run, then this will only bring in incremental data as opposed to a full pull. So we will have computers with the discovery source of 'SG-SCCM' in sys_object_source for small number of computers selected in that incremental pull and not for all computers. On the other hand, if we bring in a full pull of OS, Disk, Processor and Software which has references to computers that does not exist yet in CMDB, then this will create large number of partial payloads. Combined with the known IRE issue, this could cause a potential P2 or P1 issue resulting in a severe performance degradation.

Work around for this issue:

1. Inspect a few partial payload records with 'true' status from the cmdb_ire_partial_payloads table. Check the resource id, which is the unique identifier (Source Native Key) for a computer referenced in the partial payload. Go to the sys_object_source table and filter on Name = 'SG-SCCM' and ID = 'resourceid'. If there are no hits for the computers, then we can be fairly certain that this is a case where a full pull of computers was not performed.

2. Delete all records from the partial payloads table (cmdb_ire_partial_payloads)

3. Open up the 'SG-SCCM Computer Identity' data source and uncheck 'Use last run datetime' and save it

4. Repeat for other data sources if necessary. Reminder that software processing could take longer time if there are millions of software in the SCCM environment considering a full pull. If a full was already done, then repeating a full pull of all sources may not be necessary. Full pull of computer is sufficient in this case.

5. Execute the 'SG-SCCM Computer Identity' import schedule 

If all computers were brought in for SG-SCCM, then partial payloads will not be created in this particular case and hence avoid the IRE partial payload known issue (performance degradation) and thereby avoiding a potential P1/P2 case.