Fleet should be able to include automatically field definitions that depend on the collector, or on custom processing, and not on a package.
So as examples:
The Apache integration is installed by a user. Fleet should install in the index templates both, the mappings included in the Apache package, plus the mappings provided by the Agent Common Schema for the latest known version of Elastic Agent.
The httpjson input is configured by a user with additional custom processing. Fleet should install in the index templates the mappings included in the httpjson input package, plus the mappings of any processor used in the custom processing.
Some unknowns:
How are these mappings distributed? Are they included in the elastic_agent package? In other packages?
Can Fleet reliably detect the processors used in custom processing?
Describe a specific use case for the feature:
Support mappings of processors that cannot be known beforehand by package or Fleet developers.
This feature is blocked by the definition of https://github.com/elastic/package-spec/issues/441, and should be coordinated with https://github.com/elastic/package-spec/issues/441
Describe the feature:
Fleet should be able to include automatically field definitions that depend on the collector, or on custom processing, and not on a package.
So as examples:
Some unknowns:
Describe a specific use case for the feature:
Support mappings of processors that cannot be known beforehand by package or Fleet developers.