feat: Added New Relic Control health check

[jsumners-nr]

This PR resolves #2838.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.38%. Comparing base (e049442) to head (454c4c5).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2841      +/-   ##
==========================================
+ Coverage   88.69%   97.38%   +8.68%     
==========================================
  Files         312      314       +2     
  Lines       47635    47979     +344     
==========================================
+ Hits        42248    46722    +4474     
+ Misses       5387     1257    -4130     

Flag	Coverage Δ
integration-tests-cjs-18.x	72.40% <62.13%> (?)
integration-tests-cjs-20.x	72.40% <62.13%> (?)
integration-tests-cjs-22.x	72.62% <62.13%> (-0.08%)	⬇️
integration-tests-esm-18.x	49.65% <52.63%> (+0.04%)	⬆️
integration-tests-esm-20.x	49.66% <52.63%> (+0.04%)	⬆️
integration-tests-esm-22.x	49.71% <52.63%> (+0.04%)	⬆️
unit-tests-18.x	89.20% <100.00%> (?)
unit-tests-20.x	89.20% <100.00%> (?)
unit-tests-22.x	89.21% <100.00%> (?)
versioned-tests-18.x	79.06% <52.63%> (-0.36%)	⬇️
versioned-tests-20.x	79.07% <52.63%> (-0.36%)	⬇️
versioned-tests-22.x	79.08% <52.63%> (-0.35%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[bizob2828]

i know this is in draft, just called out the lack of config items defined

[bizob2828]

I haven't completely reviewed but a few comments. Also I don't see tests for every health status. At the very least we should have already have tests for STATUS_HEALTHY, STATUS_INVALID_LICENSE_KEY, STATUS_AGENT_DISABLED, STATUS_CONNECT_ERROR

[bizob2828]

these env overrides are not necessary. it defaults to NEW_RELIC_ path to config value

[jsumners-nr]

Are you suggesting to not provide default values? I don't understand the review comment.

[jsumners-nr]

Oh, you're saying that { agent_control: { enabled } } will already be parsed from NEW_RELIC_AGENT_CONTROL_ENABLED regardless of setting the env key. I was unaware of that. Are you strongly against setting the env key? It is clearer to me.

[bizob2828]

this key is for overrides. it'll default to what you provided so it just confuses things. we default all env vars here

[bizob2828]

I am against it because this was built to derive env vars. In fact before I built that support here we had drift and was constantly fixing one offs. The intent of the env key is to specify overrides when it doesn't fit our convention.

• an override if the env var does not follow our standard convention. We should not have

[bizob2828]

maybe we can add PR separately and rename it envOverride or something, or we could document the structure better

[jsumners-nr]

Yeah, renaming it to envVarNameOverride would be a lot clearer. I'll go ahead and remove them.

[bizob2828]

After re-reviewing spec, there are a few codes that are missing/in the wrong spot:

code	status	notes
NR-APM-000	Healthy	✅
NR-APM-001	Invalid license key (HTTP status code 401)	This occurs in lib/collector/api. You should be able to use this._agent.healthReporter to properly log these here
NR-APM-002	License key missing in configuration	✅
NR-APM-003	Forced disconnect received from New Relic (HTTP status code 410)	TThis occurs in lib/collector/api. You should be able to use this._agent.healthReporter to properly log these here
NR-APM-004	HTTP error response code [%s] received from New Relic while sending data type [%s]	I'm not sure on this one
NR-APM-005	Missing application name in agent configuration	✅
NR-APM-006	The maximum number of configured app names (3) exceeded	it does not appear we check for this
NR-APM-007	HTTP Proxy configuration error; response code [%s]	This occurs in lib/collector/api. You should be able to use this._agent.healthReporter to properly log these here
NR-APM-008	Agent is disabled via configuration	✅
NR-APM-009	Failed to connect to New Relic data collector	I'm not sure if this is getting conflated with the 410 error above
NR-APM-010	Agent config file is not able to be parsed	we're missing this: node-newrelic/index.js at main · newrelic/node-newrelic, however I don't get this because we cannot use the health reporter until the config is parsed. UPDATE: seems like we can ignore this because this case can never be hit for us
NR-APM-099	Agent has shutdown SHOULD only be reported if agent is "healthy" on shutdown	✅

[bizob2828]

Looking good. I finally did a thorough pass and called out some missing status codes or they are not in the correct place. there are also merge conflicts with main

[bizob2828]

One last test, asserting that we're setting the agent disabled status. Looks like we have or the next few subsequent tests

test('should not throw with valid config', () => {
  const config = configurator.initialize({ agent_enabled: false })
  const agent = new Agent(config)
  assert.equal(agent.config.agent_enabled, false)
})

[jsumners-nr]

One last test, asserting that we're setting the agent disabled status. Looks like we have or the next few subsequent tests

test('should not throw with valid config', () => {
  const config = configurator.initialize({ agent_enabled: false })
  const agent = new Agent(config)
  assert.equal(agent.config.agent_enabled, false)
})

We can't do that. Our "agent" doesn't get created if agent_enabled is set to false. So we don't yet have a health reporter to update:

node-newrelic/index.js

Lines 92 to 106 in 6cde2e4

	if (!config) {
	logger.info('No configuration detected. Not starting.')
	} else if (!config.agent_enabled) {
	logger.info('Module disabled in configuration. Not starting.')
	} else if (!config.worker_threads.enabled && !isMainThread) {
	logger.warn(
	'New Relic for Node.js in worker_threads is not officially supported. Not starting! To bypass this, set `config.worker_threads.enabled` to true in configuration.'
	)
	} else {
	if (!isMainThread && config.worker_threads.enabled) {
	logger.warn(
	'Attempting to load agent in worker thread. This is not officially supported. Use at your own risk.'
	)
	}
	agent = createAgent(config)

---

But I updated the test to cover the lines in the agent start. It's just never going to actually matter.