[Agent] Configuration validations on Fleet and Agent side

[ph]

We allow people to see and edit the configuration in Fleet, it might be a good idea to share validations or at least high-level validations with the expected fields like ids, type and metricset. This can be also useful for communicating with other teams.

We have created an example configuration but this is not enough to express what we are expecting. We need a formal way to define it, one way would be to use a json-schema definition that could be used by both the agent and fleet.

[elasticmachine]

Pinging @elastic/ingest (Project:fleet)

[elasticmachine]

Pinging @elastic/ingest-management (Team:ingest-management)

[elasticmachine]

Pinging @elastic/elastic-agent-control-plane (Team:Elastic-Agent-Control-Plane)

[jlind23]

@nimarezainia Relevant from a customer experience?

[jlind23]

@ph Were you thinking of adding those validation inside fleet ui?

[ph]

I believe it's a better user experience to catch user error as early as possible. This also has the adventages to reduce load on fleet-server and on the user network when a configuration is invalid.

I haven't investigated but, integration generates input and we can validate inputs at the process level maybe we can share the heuristic between our inputs and fleet to validate it?

The other way would be to take a random agent and ask to validate the configuration.