sailfork / t3code-github / .repos/ alchemy-effect/ packages/ integration-tests/ test/ Cloudflare/ Container/ ContainerRestart.test.ts

import * as Cloudflare from "@alchemy.run/cloudflare";
import { describe, expect } from "@alchemy.run/test";
import * as Test from "@alchemy.run/test/Alchemy";
import * as Effect from "effect/Effect";
import * as HttpClient from "effect/http/HttpClient";
import * as HttpClientRequest from "effect/http/HttpClientRequest";
import { MinimumLogLevel } from "effect/References";
import * as Schedule from "effect/Schedule";
import RestartStack from "./fixtures/restart/stack.ts";

const { test, beforeAll, afterAll, deploy, destroy } = Test.make({
  providers: Cloudflare.providers(),
  state: Cloudflare.state(),
});

const logLevel = Effect.provideService(MinimumLogLevel, process.env.DEBUG ? "Debug" : "Info");

// Image build + push + worker/DO deploy comfortably exceeds the default hook
// budget, and a restart adds another cold start on top, so be generous.
const HOOK_TIMEOUT = 600_000;
const TEST_TIMEOUT = 300_000;

const DEPLOY_PLACEHOLDER = "Alchemy worker is being deployed...";

// Force `Connection: close` so each attempt opens a fresh connection and can
// land on an edge that already has the new deploy / restarted container.
const freshConn = HttpClient.mapRequest(HttpClientRequest.setHeader("connection", "close"));

const readinessSchedule = Schedule.min([
  Schedule.exponential("500 millis"),
  Schedule.spaced("3 seconds"),
]);

// Retry a route until it answers 200 with a body containing `expected`,
// rejecting transient non-200s and the pre-create deploy stub. This is how we
// assert "the container is up and serving" both initially and after a restart.
const fetchReady = (url: string, expected: string) =>
  Effect.gen(function* () {
    const client = freshConn(yield* HttpClient.HttpClient);
    return yield* client.get(url).pipe(
      Effect.flatMap((r) =>
        r.status !== 200
          ? Effect.fail(new Error(`not ready: ${r.status}`))
          : Effect.flatMap(r.text, (body) =>
              body.includes(DEPLOY_PLACEHOLDER) || !body.includes(expected)
                ? Effect.fail(new Error(`not ready: got ${body}`))
                : Effect.succeed(body),
            ),
      ),
      Effect.timeout("30 seconds"),
      Effect.retry({ schedule: readinessSchedule, times: 40 }),
    );
  });

// Fire a single GET and return its body (used for /stop, /crash).
const hit = (url: string) =>
  Effect.gen(function* () {
    const client = freshConn(yield* HttpClient.HttpClient);
    const res = yield* client.get(url);
    return yield* res.text;
  });

// Poll `/running` until the reported state matches `want`. Used to confirm the
// container actually went down before we assert the next request restarts it,
// so the test exercises the real restart path (not a still-warm container).
const waitRunning = (baseUrl: string, name: string, want: boolean) =>
  Effect.gen(function* () {
    const client = freshConn(yield* HttpClient.HttpClient);
    return yield* client.get(`${baseUrl}/running?name=${encodeURIComponent(name)}`).pipe(
      Effect.flatMap((r) =>
        r.status !== 200
          ? Effect.fail(new Error(`running not ready: ${r.status}`))
          : Effect.flatMap(r.json, (body) => {
              const running = (body as { running?: boolean }).running;
              return running === want
                ? Effect.succeed(running)
                : Effect.fail(new Error(`running=${running}, want ${want}`));
            }),
      ),
      Effect.timeout("15 seconds"),
      Effect.retry({ schedule: readinessSchedule, times: 40 }),
    );
  });

/**
 * Auto-restart: a Durable Object's container that has stopped or crashed must
 * be transparently restarted on the next request. Regression coverage for the
 * `readyPorts` cache going stale — previously a port confirmed ready once was
 * never re-probed and `ensureRunning` was skipped, so a stopped container was
 * never restarted.
 */
describe(
  "container auto-restart",
  {
    tags: [
      "provider:cloudflare",
      "provider:cloudflare:container",
      "provider:cloudflare:worker",
      "live",
    ],
  },
  () => {
    const stack = beforeAll(deploy(RestartStack), { timeout: HOOK_TIMEOUT });
    afterAll.skipIf(!!process.env.NO_DESTROY)(destroy(RestartStack), { timeout: HOOK_TIMEOUT });

    test(
      "restarts the container after it is stopped (destroy)",
      Effect.gen(function* () {
        const { url } = yield* stack;
        const name = "stop";

        // Start + confirm up.
        expect(yield* fetchReady(`${url}/ping?name=${name}`, "pong")).toContain("pong");

        // Hard-stop it, then confirm it is actually down before re-pinging — so
        // the next ping must go through the restart path.
        yield* hit(`${url}/stop?name=${name}`);
        yield* waitRunning(url, name, false);

        // Next request transparently restarts it.
        expect(yield* fetchReady(`${url}/ping?name=${name}`, "pong")).toContain("pong");
        expect(yield* waitRunning(url, name, true)).toBe(true);
      }).pipe(logLevel),
      { timeout: TEST_TIMEOUT },
    );

    test(
      "restarts the container after it crashes (non-zero exit)",
      Effect.gen(function* () {
        const { url } = yield* stack;
        const name = "crash";

        expect(yield* fetchReady(`${url}/ping?name=${name}`, "pong")).toContain("pong");

        // Make the container process exit on its own, then wait for the monitor
        // to observe the exit (running === false).
        yield* hit(`${url}/crash?name=${name}`);
        yield* waitRunning(url, name, false);

        // Next request transparently restarts it.
        expect(yield* fetchReady(`${url}/ping?name=${name}`, "pong")).toContain("pong");
        expect(yield* waitRunning(url, name, true)).toBe(true);
      }).pipe(logLevel),
      { timeout: TEST_TIMEOUT },
    );
  },
);