Elasticsearch aggregation on multiple fields across multiple indexes
Fields can only be aggregated across indices if they are named alike. There is no wildcard syntax for aggregation fields.
Here is what your mapping currently defines:
INDEX: company-company_databases
TYPE: company_database
FIELD NAMES:
company_applications.company_system_applications.vendor_name
company_applications.company_system_applications.system_application.vendor_name
INDEX: applications
TYPE: application
FIELD NAMES:
company_databases.company_applications.company_system_applications.vendor_name
company_databases.company_applications.company_system_applications.system_application.vendor_name
As far as Elasticsearch is concerned, these fields have nothing in common (even though part of the path is vendor_name
).
If your goal is to aggregate vendor_name
across a query that spans the two indices, think about restructuring your indices/mappings to accomplish this.
Note that Elasticsearch doesn't model many-to-many relationships
If you can get away with duplicating Database info across applications, you might be able to re-formulate your relationships as a hierarchy, e.g.:
INDEX: applications
--
TYPE: application
FIELDS: vendor_name
, etc...
--
TYPE: database_application
FIELDS: vendor_name
, databases.<inner fields>
, etc...
--
Then you'd be able to aggregate across types on the same field path vendor_name
with the added bonus of querying a single applications
index.