diff --git a/docs/tutorial/IndexAugmenter.md b/docs/tutorial/IndexAugmenter.md index fc4d8dc2751b58feffb77ea36560b617ef21d6bc..48fcfa97ba8f6060be8f3e26332f921e60496fc0 100644 --- a/docs/tutorial/IndexAugmenter.md +++ b/docs/tutorial/IndexAugmenter.md @@ -324,11 +324,15 @@ the records from the `OSDU search` results, any one of the following mistakes ca * The feature flag `index-augmenter-enabled` for `Index Augmenter` is not enabled in the given data partition. Please check with the service provider. + * If the extended properties are created in the data records but those extended properties are not searchable, the option `force_clean` for re-index is not set to `True`. + * Any one of the mandatory properties is missing, such as `data.Code`, `data.Configurations[].Name`, `data.Configurations[].Policy` or `data.Configurations[].Paths[].ValueExtraction.ValuePath` and etc. + * The value of `data.Code` in the record is not a major schema version kind which is ended with version major and dot. + * Multiple IndexPropertyPathConfiguration records to a major schema version kind may exist. Using kind `osdu:wks:work-product-component--WellLog:1.` as example to run OSDU query: ``` @@ -337,6 +341,7 @@ the records from the `OSDU search` results, any one of the following mistakes ca "query": "data.Code: "\osdu:wks:work-product-component--WellLog:1."\" } ``` + * If not all extended properties are missing, the `Configurations[]` of the missing extended properties could be invalid. The `Index Augmenter` engine can do basic syntax check on each configuration and only ignore the invalid ones.