Issue Summary

One JSON import synchronization generates duplicated objects, even though Identifier is unique.

Steps to Reproduce

1. This is not reproducible in house - only in one customer's environments.
2. A Jason file is being generated daily, accessed by the Insight JSON Import, using a URL. File is being generated in advance, way before the import initiates.
3. Almost every run will create new objects, some of these objects will be created twice, always in the same 3 object types.
4. Duplicated objects seems to be random, with no pattern, not always created in all 3 object types.
5. Identifier is unique - tested with various combinations of identifiers, ended up with single datalocators mapped to text attributes, and the issue persists.
6. Recreated the import configuration, made sure no rogue/corrupt configuration exists in the DB.
7. Recreated the object types from scratch, and their attributes to no avail.
8. The results are random duplicated objects, created within 15m apart.
9. The youngest instance will get ALL its outbound references set, while the older instance will have some but not all, always the same reference attributes are missing. (An attribute is mapped with 3 datalocators, all combined in one IQL query on the attribute map.
10. Multiple attempts to reproduce this behavior in house, using JSON file from the customer's instance, and replicating the Structure (object types involved) - resulted without success. The import works as expected and no duplicates are being created.

Expected Results

No duplicate objects are present following the completion of the JSON import

Actual Results

The below exception is thrown in the xxxxxxx.log file:

...

Workaround

Possible workaround, as per the customer experiencing it:
In the Import>Object Type Maps>Reference Attribute configuration they had IQL mapping with dot notations.
Once they have replaced the dot notation IQL with outR() no more duplicates were created!
e.g. instead of having:
Name = ${datalocator1} AND AttributeX.Name=${datalocator2} AND AttributeY=${datalocator3}
use the following IQL:
Name = ${datalocator1} AND object HAVING outR(Name=${datalocator2},reftype IN (myRefType)) AND AttributeY=${datalocator3}

Note: Since these Attributes are NOT set to be identifiers, this should not have an effect on the creation of Duplicated objects, as References are being added to Objects as the last step of the Import process. BUT - adjusting the IQL did resolve the issue for the customer experiencing this!