Introduced 'Pending' state for backend's health

[andythsu]

This commit adds a 'Pending' state to keep track of backend's health.

Resolves: #80

[cla-bot]

Thank you for your pull request and welcome to our community. We could not parse the GitHub identity of the following contributors: Andy Su (Apps).
This is most likely caused by a git client misconfiguration; please make sure to:

1. check if your git client is configured with an email to sign commits git config --list | grep email
2. If not, set it up using git config --global user.email [email protected]
3. Make sure that the git commit email is configured in your GitHub account settings, see https://github.com/settings/emails

[rdsarvar]

nitpick: because you're modifying the method signature we might as well fix the typo and consistency of BackEnd vs Backend: upateBackEndHealth -> updateBackendHealth

(feel free to resolve with or without the fix, i'm poking my head in here and might not see the response :D)

[rdsarvar]

nitpick: potentially use jakarta.ws.rs.core.Response.Status.NOT_FOUND.getStatusCode() or HttpURLConnection.HTTP_NOT_FOUND (<-- preference) here instead of the originally magic number?

(note: feel free to resolve with or without making the modifications)

[rdsarvar]

nitpick: perhaps here we want to hit all enum states in the test?

for ex: 50% pending and 50% unhealthy which both represent "not yet healthy"
-> i % 2 == 0 ? BackendHealthState.UNHEALTHY : BackendHealthState.PENDING

(note: adding this note to all blocks as i might not be able to get back to them; feel free to resolve with or without change)

[rdsarvar]

question: is the goal of this test case to be a sub-system test?

my thinking is: if we just wanted to test that the gateway chooses backends properly based on health as a single unit you could replace:

    SECONDS.sleep(WAIT_FOR_BACKEND_IN_SECONDS);

with:

    HaGatewayLauncher haGatewayLauncher = new HaGatewayLauncher("io.trino");
    haGatewayLauncher.run(args);

    // Now populate the backend
    setUpBackend(
            "trino1", "http://localhost:" + backendPort, "externalUrl", true, "adhoc", routerPort);

    // When a backend becomes active for the first time OR is transitioned from inactive to active it will not yet be
    // treated as healthy. Without a real health check happening we will extract the RoutingManager and set them to healthy manually
    Field privateInjectorField = BaseApp.class.
            getDeclaredField("injector");
    privateInjectorField.setAccessible(true);

    Injector injector = (Injector) privateInjectorField.get(haGatewayLauncher);
    injector.getInstance(RoutingManager.class).updateBackendHealth("trino1", BackendHealth.HEALTHY);

This hack to grab the class with reflection is just to keep the test as close to the module as possible (and do without a sleep pause). Curious on your thoughts for it.

If the goal is to cover as much repository as possible in the test as sub-system instead of integration then fair enough 👍

[andythsu]

great idea! included this in my new PR