feat(platform): show which startup step is running, with progress and ETA (#3813)

Co-authored-by: nova28 <17953305+nova28@users.noreply.github.com> Co-authored-by: Carlos Florencio <carlosmflorencio@gmail.com>
2026-09-20 07:06:19 +08:00
parent 1727435c14
commit 8779c88d79
124 changed files with 7527 additions and 347 deletions
@@ -0,0 +1,58 @@
+import { expect, test } from "../../fixtures/test-base";
+import {
+  setAgentRuntimeAvailability,
+  stubAgentRuntimeRestart,
+} from "../../helpers/agent-runtime-availability";
+import { assertNoDocumentHorizontalOverflow } from "../../helpers/layout-assertions";
+
+const PROGRESS_BODY = {
+  status: "starting",
+  startup: {
+    phase: "applying_migrations",
+    boot: 1,
+    seq: 1,
+    elapsed_ms: 12_000,
+    phase_elapsed_ms: 12_000,
+    step: {
+      id: "stores.services",
+      label_key: "startup.step.stores_services",
+      measure: "counted",
+      unit: "messages",
+      elapsed_ms: 12_000,
+      done: 400,
+      total: 1000,
+      eta_ms: 30_000,
+      stalled: false,
+    },
+  },
+};
+
+test("renders startup progress in the mobile restart dialog", async ({ testPage }) => {
+  await stubAgentRuntimeRestart(testPage);
+  await testPage.route("**/ready", (route) =>
+    route.fulfill({
+      status: 503,
+      contentType: "application/json",
+      body: JSON.stringify(PROGRESS_BODY),
+    }),
+  );
+
+  await testPage.goto("/stats");
+  await expect(testPage.getByTestId("app-shell")).toBeVisible();
+
+  await setAgentRuntimeAvailability(testPage, {
+    status: "unavailable",
+    reason: "agentctl_exited",
+    occurred_at: "2026-09-18T00:00:00Z",
+  });
+  const alert = testPage.getByTestId("agent-runtime-alert");
+  await expect(alert).toBeVisible();
+  await alert.getByRole("button", { name: "Restart Kandev" }).tap();
+
+  const dialog = testPage.getByTestId("restart-progress-dialog");
+  await expect(dialog).toHaveAttribute("data-phase", "restarting");
+  await expect(dialog).toContainText("Service stores");
+  await expect(dialog).toContainText("400 of 1000 messages");
+  await expect(dialog.getByRole("progressbar")).toHaveAttribute("aria-valuenow", "40");
+  await assertNoDocumentHorizontalOverflow(testPage, "mobile startup progress dialog");
+});
@@ -0,0 +1,184 @@
+import { expect, test } from "../../fixtures/test-base";
+import {
+  setAgentRuntimeAvailability,
+  stubAgentRuntimeRestart,
+} from "../../helpers/agent-runtime-availability";
+
+type ReadyMode = "progress" | "unreachable" | "ready";
+
+const PROGRESS_BODY = {
+  service: "kandev",
+  version: "e2e-test",
+  status: "starting",
+  startup: {
+    phase: "applying_migrations",
+    boot: 1,
+    seq: 1,
+    elapsed_ms: 12_000,
+    phase_elapsed_ms: 12_000,
+    step: {
+      id: "stores.services",
+      label_key: "startup.step.stores_services",
+      measure: "counted",
+      unit: "messages",
+      elapsed_ms: 12_000,
+      done: 400,
+      total: 1000,
+      rate_per_second: 33.3,
+      eta_ms: 30_000,
+      since_advance_ms: 500,
+      stalled: false,
+    },
+  },
+};
+
+const UNREACHABLE_BODY = { error: "internal error" };
+
+const READY_BODY = { service: "kandev", version: "e2e-test", status: "ok" };
+
+const STALLED_BODY = {
+  service: "kandev",
+  version: "e2e-test",
+  status: "starting",
+  startup: {
+    phase: "applying_migrations",
+    boot: 1,
+    seq: 1,
+    elapsed_ms: 130_000,
+    phase_elapsed_ms: 130_000,
+    step: {
+      id: "stores.services",
+      label_key: "startup.step.stores_services",
+      measure: "counted",
+      unit: "messages",
+      elapsed_ms: 130_000,
+      done: 400,
+      total: 1000,
+      since_advance_ms: 125_000,
+      stalled: true,
+    },
+  },
+};
+
+/**
+ * Drives RestartProgressDialog through a real (fully stubbed) restart and
+ * asserts the step-level detail added on top of the base phase/elapsed
+ * rendering (AC-PLATFORM-STARTUP-PROGRESS-003): step label, counted
+ * progress text, ETA, and the AC-PLATFORM-STARTUP-PROGRESS-002.6 status-code
+ * fallback rule for a response with no parseable snapshot - a failing status
+ * falls back to last-known, a successful one does not.
+ */
+test.describe("Restart progress dialog startup detail", () => {
+  test("renders step label and progress, and resolves a snapshot-free response by status", async ({
+    testPage,
+  }) => {
+    test.setTimeout(60_000);
+
+    let readyMode: ReadyMode = "progress";
+    await stubAgentRuntimeRestart(testPage);
+    await testPage.route("**/ready", (route) => {
+      if (readyMode === "progress") {
+        return route.fulfill({
+          status: 503,
+          contentType: "application/json",
+          body: JSON.stringify(PROGRESS_BODY),
+        });
+      }
+      if (readyMode === "unreachable") {
+        return route.fulfill({
+          status: 500,
+          contentType: "application/json",
+          body: JSON.stringify(UNREACHABLE_BODY),
+        });
+      }
+      return route.fulfill({
+        status: 200,
+        contentType: "application/json",
+        body: JSON.stringify(READY_BODY),
+      });
+    });
+
+    await testPage.goto("/stats");
+    await expect(testPage.getByTestId("app-shell")).toBeVisible();
+
+    await setAgentRuntimeAvailability(testPage, {
+      status: "unavailable",
+      reason: "agentctl_exited",
+      occurred_at: "2026-09-18T00:00:00Z",
+    });
+    const alert = testPage.getByTestId("agent-runtime-alert");
+    await expect(alert).toBeVisible();
+    await alert.getByRole("button", { name: "Restart Kandev" }).click();
+
+    const dialog = testPage.getByTestId("restart-progress-dialog");
+    await expect(dialog).toHaveAttribute("data-phase", "restarting");
+
+    await expect(dialog).toContainText("Applying migrations");
+    await expect(dialog).toContainText("Service stores");
+    await expect(dialog).toContainText("400 of 1000 messages");
+    await expect(dialog).toContainText("About 30 seconds remaining");
+    await expect(dialog.getByRole("progressbar")).toHaveAttribute("aria-valuenow", "40");
+    await expect(dialog.getByText("Last known")).toHaveCount(0);
+
+    // A failing status with no parseable snapshot falls back to last-known
+    // (AC-PLATFORM-STARTUP-PROGRESS-002.6/003.13): the prior step detail
+    // stays on screen, labelled stale rather than cleared.
+    readyMode = "unreachable";
+
+    await expect(dialog.getByText("Last known")).toBeVisible();
+    await expect(dialog).toContainText("Service stores");
+    await expect(dialog).toContainText("400 of 1000 messages");
+
+    // A successful status with no parseable snapshot means ready, not stale
+    // (AC-PLATFORM-STARTUP-PROGRESS-002.6): the status code is authoritative,
+    // so "Last known" must clear even though this response has no snapshot
+    // either.
+    readyMode = "ready";
+
+    await expect(dialog.getByText("Last known")).toHaveCount(0);
+  });
+
+  /**
+   * Covers AC-PLATFORM-STARTUP-PROGRESS-004.6: the Settings restart dialog
+   * must distinguish a stalled step from a running one, state how long it has
+   * gone without progress, and clear that state within one refresh interval
+   * once the step advances again.
+   */
+  test("shows a stalled step and clears it once progress resumes", async ({ testPage }) => {
+    test.setTimeout(60_000);
+
+    let stalled = true;
+    await stubAgentRuntimeRestart(testPage);
+    await testPage.route("**/ready", (route) => {
+      const body = stalled ? STALLED_BODY : PROGRESS_BODY;
+      return route.fulfill({
+        status: 503,
+        contentType: "application/json",
+        body: JSON.stringify(body),
+      });
+    });
+
+    await testPage.goto("/stats");
+    await expect(testPage.getByTestId("app-shell")).toBeVisible();
+
+    await setAgentRuntimeAvailability(testPage, {
+      status: "unavailable",
+      reason: "agentctl_exited",
+      occurred_at: "2026-09-18T00:00:00Z",
+    });
+    const alert = testPage.getByTestId("agent-runtime-alert");
+    await expect(alert).toBeVisible();
+    await alert.getByRole("button", { name: "Restart Kandev" }).click();
+
+    const dialog = testPage.getByTestId("restart-progress-dialog");
+    await expect(dialog).toHaveAttribute("data-phase", "restarting");
+
+    await expect(dialog).toContainText("400 of 1000 messages");
+    await expect(dialog).toContainText("Stalled for 3 minutes");
+
+    stalled = false;
+
+    await expect(dialog).not.toContainText("Stalled for 3 minutes");
+    await expect(dialog).toContainText("400 of 1000 messages");
+  });
+});
@@ -0,0 +1,116 @@
+import type { Route } from "@playwright/test";
+import { expect, test } from "../../fixtures/test-base";
+
+type StartupSnapshot = Record<string, unknown>;
+
+const STARTING_SNAPSHOT: StartupSnapshot = {
+  phase: "applying_migrations",
+  boot: 1,
+  seq: 1,
+  elapsed_ms: 12_000,
+  phase_elapsed_ms: 12_000,
+  step: {
+    id: "stores.services",
+    label_key: "startup.step.stores_services",
+    measure: "counted",
+    unit: "messages",
+    elapsed_ms: 12_000,
+    done: 400,
+    total: 1000,
+    eta_ms: 30_000,
+    stalled: false,
+  },
+};
+
+const ADVANCED_BODY = {
+  status: "starting",
+  startup: {
+    ...STARTING_SNAPSHOT,
+    seq: 2,
+    step: { ...(STARTING_SNAPSHOT.step as Record<string, unknown>), done: 700, eta_ms: 10_000 },
+  },
+};
+
+const READY_BODY = {
+  status: "ok",
+  startup: { phase: "ready", boot: 1, seq: 3, elapsed_ms: 20_000, phase_elapsed_ms: 500 },
+};
+
+const POST_RELOAD_HTML =
+  '<!DOCTYPE html><html><body><div data-testid="post-reload-marker">app is up</div></body></html>';
+
+/**
+ * Drives the Go-rendered startup page (internal/backendapp/startup_page.go)
+ * through a real browser instead of httptest string-matching: the initial
+ * markup comes from the production renderer via the E2E-only fixture route,
+ * and only /ready and the app navigation are mocked, so the embedded poll
+ * script runs unmodified.
+ */
+test.describe("Startup page (real render)", () => {
+  test("polls for step progress and reloads once the snapshot reports ready", async ({
+    backend,
+    browser,
+    request,
+  }) => {
+    const startingHtml = await request
+      .post(`${backend.baseUrl}/api/v1/e2e/startup-page-fixture`, { data: STARTING_SNAPSHOT })
+      .then((res) => res.text());
+
+    const context = await browser.newContext({ baseURL: backend.frontendUrl });
+    const page = await context.newPage();
+
+    let readyBody: unknown = { status: "starting", startup: STARTING_SNAPSHOT };
+    let appIsReady = false;
+
+    const fulfillReady = (route: Route) =>
+      route.fulfill({
+        status: appIsReady ? 200 : 503,
+        contentType: "application/json",
+        body: JSON.stringify(readyBody),
+      });
+
+    const fulfillAppRoute = (route: Route) =>
+      appIsReady
+        ? route.fulfill({ status: 200, contentType: "text/html", body: POST_RELOAD_HTML })
+        : route.fulfill({ status: 503, contentType: "text/html", body: startingHtml });
+
+    await page.route(
+      (url) => url.pathname === "/ready",
+      (route) => fulfillReady(route),
+    );
+    await page.route(
+      (url) => url.pathname === "/startup-page-e2e-fixture",
+      (route) => fulfillAppRoute(route),
+    );
+
+    await page.goto("/startup-page-e2e-fixture");
+
+    await expect(page.locator("#startup-phase")).toHaveText("Applying migrations");
+    await expect(page.locator("#startup-step-label")).toHaveText("Service stores");
+    await expect(page.locator("#startup-progress")).toHaveText("400 of 1000 messages");
+    await expect(page.locator("#startup-eta")).toHaveText("About 30 seconds remaining");
+
+    // A failing response with a malformed snapshot must keep the last valid
+    // detail and mark it as stale. The status code remains authoritative when
+    // the body shape cannot be trusted.
+    readyBody = { status: "starting", startup: { ...STARTING_SNAPSHOT, step: {} } };
+    await expect(page.getByText("Last known")).toBeVisible();
+    await expect(page.locator("#startup-step-label")).toHaveText("Service stores");
+    await expect(page.locator("#startup-progress")).toHaveText("400 of 1000 messages");
+
+    // Poll-driven update: the embedded script re-fetches /ready on its own
+    // schedule and re-renders in place, with no test-issued reload.
+    readyBody = ADVANCED_BODY;
+    await expect(page.locator("#startup-progress")).toHaveText("700 of 1000 messages");
+    await expect(page.locator("#startup-eta")).toHaveText("About 10 seconds remaining");
+
+    // Reload-on-ready: once a poll response's snapshot reports phase "ready",
+    // the script calls location.reload() itself; the re-navigation this test
+    // observes is the real app route being requested a second time.
+    appIsReady = true;
+    readyBody = READY_BODY;
+    await expect(page.getByTestId("post-reload-marker")).toBeVisible();
+
+    await context.close();
+  });
+});