[Fleet] Telemetry for Fleet Server

[mostlyjason]

We'd like to track telemetry on Fleet Server so we can better understand if users are successful in setting it up, where they find difficulty with it, and what features they are using.

Our existing telemetry can already tell us:

• Whether Fleet Server is running
• How many agents are enrolled
• Whether the Fleet Server integration is enabled (automatically enabled when a superuser visits Fleet)

We'd like to add stats on the following:

• Total number of Elastic Agents with Fleet Server, including offline and inactive ones. This will help us identify if a cluster ever successfully got Fleet Server running, even if they are no longer using it or it is unenrolled.
• Number of healthy Fleet Servers. This will tell us which clusters currently have a running Fleet server. It also help us determine how common HA and/or load balanced deployments are.

Stretch:

• Include the other statuses for Elastic Agent, but only counting ones with Fleet Server All UI Agent state options are not displayed in Telemetry collection data #86932 (comment)
• Number of Fleet Server host URLs. If we see a significant number of self-managed clusters with a host and no Fleet Servers, it may indicate difficulty setting it up.

Schema:

"fleet" : {
  "fleet_server": {
    "total_all_statuses": 3,
    "healthy": 1,
    "num_host_urls": 1
  }
}

Related:

• Please use the same field definitions from All UI Agent state options are not displayed in Telemetry collection data #86932 (comment)

[elasticmachine]

Pinging @elastic/fleet (Team:Fleet)

[skh]

In general, everything the kibana user can read can also be reported as telemetry. @nchaulet can you comment on the additional stats for fleet server?

[nchaulet]

Include the other statuses for Elastic Agent, but only counting ones with Fleet Server #86932 (comment)

@mostlyjason can you clarify what is missing here?

[mostlyjason]

@nchaulet we don't have status data broken out for fleet servers today:

"fleet" : {
  "fleet_server": {
    "total_all_statuses": 3,
    "healthy": 1,
    "num_host_urls": 1
  }
}

[nchaulet]

@mostlyjason do we want to have the status broken out? to have something similar to the agents telemetry data?

[mostlyjason]

@nchaulet this is closed, but don't we need to update the mappings int he telemetry cluster too?

[nchaulet]

Thanks for catching this @mostlyjason just create the PR for the mappings in the telemetry cluster here https://github.com/elastic/telemetry/issues/538