@oneuptime/common 14.0.2 → 14.0.5

package/Tests/App/Dashboard/CloudMetricTooltips.test.tsx

@@ -0,0 +1,887 @@
+import {
+  afterEach,
+  beforeEach,
+  describe,
+  expect,
+  jest,
+  test,
+} from "@jest/globals";
+import "@testing-library/jest-dom";
+import {
+  act,
+  cleanup,
+  fireEvent,
+  render,
+  screen,
+  within,
+} from "@testing-library/react";
+import * as React from "react";
+import { MemoryRouter } from "react-router-dom";
+
+/*
+ * The (i) tooltips on the Cloud pages: the environment overview (six
+ * tiles, two charts and the "Top instances by CPU" card) and the fleet
+ * strip above the Cloud Environments list. The texts are held to the
+ * shared rules and to what the pages compute - two clocks on one overview,
+ * a live snapshot and the selected range - and the pages are rendered for
+ * real over a fake API to show each text beside the right title.
+ *
+ * jest.mock is hoisted above the imports, so the factories only close over
+ * names that start with "mock".
+ */
+
+interface MockAggregateCall {
+  modelType: { name: string };
+  aggregateBy: {
+    aggregationType: string;
+    query: Record<string, unknown>;
+  };
+}
+
+interface MockCountCall {
+  modelType: { name: string };
+  query: Record<string, unknown>;
+}
+
+interface MockInstanceRow {
+  instanceName: string;
+  latestCpuPercent?: number | undefined;
+  latestMemoryBytes?: number | undefined;
+  // Minutes before now the instance was last seen.
+  minutesAgo: number;
+}
+
+interface MockListCall {
+  modelType: { name: string };
+  sort?: Record<string, unknown> | undefined;
+}
+
+const mockAggregateCalls: Array<MockAggregateCall> = [];
+const mockListCalls: Array<MockListCall> = [];
+const mockTranslations: Record<string, string> = {};
+
+const MOCK_MIB: number = 1024 * 1024;
+
+/*
+ * The instance rows getList returns, in the order the server sends them
+ * (it sorts by latestCpuPercent descending). Reset before every test.
+ */
+const MOCK_DEFAULT_INSTANCES: Array<MockInstanceRow> = [
+  {
+    instanceName: "task-a",
+    latestCpuPercent: 80,
+    latestMemoryBytes: 512 * MOCK_MIB,
+    minutesAgo: 1,
+  },
+  {
+    instanceName: "task-b",
+    latestCpuPercent: 20,
+    latestMemoryBytes: 256 * MOCK_MIB,
+    minutesAgo: 2,
+  },
+  {
+    // Stale: outside the live window, so in no tile and no row.
+    instanceName: "task-gone",
+    latestCpuPercent: 99,
+    latestMemoryBytes: 4096 * MOCK_MIB,
+    minutesAgo: 60,
+  },
+];
+let mockInstances: Array<MockInstanceRow> = [...MOCK_DEFAULT_INSTANCES];
+// Keep the span / metric fetches (or the instance list) pending forever.
+let mockHoldAggregates: boolean = false;
+let mockHoldInstances: boolean = false;
+
+jest.mock("../../../UI/Utils/AnalyticsModelAPI/AnalyticsModelAPI", () => {
+  return {
+    __esModule: true,
+    default: {
+      aggregate: (call: MockAggregateCall): Promise<unknown> => {
+        mockAggregateCalls.push(call);
+        if (mockHoldAggregates) {
+          return new Promise<unknown>(() => {});
+        }
+        const at: Date = new Date("2026-09-24T10:00:00.000Z");
+
+        if (call.modelType.name === "Metric") {
+          return Promise.resolve({
+            data: [{ timestamp: at, value: 900 * MOCK_MIB }],
+          });
+        }
+        if (call.aggregateBy.aggregationType === "P95") {
+          return Promise.resolve({ data: [{ timestamp: at, value: 2.5e8 }] });
+        }
+        if (call.aggregateBy.query["statusCode"] !== undefined) {
+          return Promise.resolve({ data: [{ timestamp: at, value: 3 }] });
+        }
+        return Promise.resolve({ data: [{ timestamp: at, value: 100 }] });
+      },
+    },
+  };
+});
+
+jest.mock("../../../UI/Utils/ModelAPI/ModelAPI", () => {
+  return {
+    __esModule: true,
+    default: {
+      getItem: (): Promise<unknown> => {
+        return Promise.resolve({
+          name: "prod · us-east-1",
+          resourceIdentifier: "aws_ecs|123456789012|us-east-1",
+          otelCollectorStatus: "connected",
+          cloudPlatform: "aws_ecs",
+          cloudProvider: "aws",
+          cloudRegion: "us-east-1",
+          cloudAccountId: "123456789012",
+        });
+      },
+      getList: (call: MockListCall): Promise<unknown> => {
+        mockListCalls.push(call);
+        if (mockHoldInstances) {
+          return new Promise<unknown>(() => {});
+        }
+        const now: number = Date.now();
+        return Promise.resolve({
+          data: mockInstances.map(
+            (row: MockInstanceRow): Record<string, unknown> => {
+              return {
+                instanceName: row.instanceName,
+                latestCpuPercent: row.latestCpuPercent,
+                latestMemoryBytes: row.latestMemoryBytes,
+                lastSeenAt: new Date(now - row.minutesAgo * 60 * 1000),
+              };
+            },
+          ),
+        });
+      },
+      count: (call: MockCountCall): Promise<number> => {
+        if (call.modelType.name === "CloudResourceInstance") {
+          return Promise.resolve(12);
+        }
+        if (call.query["otelCollectorStatus"] === "connected") {
+          return Promise.resolve(3);
+        }
+        if (call.query["cloudProvider"] === "aws") {
+          return Promise.resolve(4);
+        }
+        if (call.query["cloudProvider"] === "gcp") {
+          return Promise.resolve(1);
+        }
+        if (typeof call.query["cloudProvider"] === "string") {
+          return Promise.resolve(0);
+        }
+        return Promise.resolve(5);
+      },
+    },
+  };
+});
+
+jest.mock("../../../UI/Utils/Project", () => {
+  const ObjectIDModule: { default: new (id: string) => unknown } =
+    jest.requireActual("../../../Types/ObjectID") as {
+      default: new (id: string) => unknown;
+    };
+  return {
+    __esModule: true,
+    default: {
+      getCurrentProjectId: (): unknown => {
+        return new ObjectIDModule.default(
+          "aaaaaaaa-0000-4000-8000-000000000001",
+        );
+      },
+    },
+  };
+});
+
+jest.mock("../../../UI/Utils/Navigation", () => {
+  const ObjectIDModule: { default: new (id: string) => unknown } =
+    jest.requireActual("../../../Types/ObjectID") as {
+      default: new (id: string) => unknown;
+    };
+  return {
+    __esModule: true,
+    default: {
+      getLastParamAsObjectID: (): unknown => {
+        return new ObjectIDModule.default(
+          "cccccccc-0000-4000-8000-000000000003",
+        );
+      },
+    },
+  };
+});
+
+jest.mock("../../../UI/Utils/API/API", () => {
+  return {
+    __esModule: true,
+    default: {
+      getFriendlyMessage: (): string => {
+        return "failed";
+      },
+    },
+  };
+});
+
+jest.mock("react-i18next", () => {
+  return {
+    useTranslation: () => {
+      return {
+        t: (value: string): string => {
+          return mockTranslations[value] ?? value;
+        },
+      };
+    },
+  };
+});
+
+jest.mock("../../../UI/Components/Charts/Line/LineChart", () => {
+  return {
+    __esModule: true,
+    default: (): React.ReactElement => {
+      return React.createElement("div", { "data-testid": "line-chart" });
+    },
+  };
+});
+
+jest.mock(
+  "../../../UI/Components/TelemetryViewer/components/TelemetryTimeRangePicker",
+  () => {
+    return {
+      __esModule: true,
+      default: (): null => {
+        return null;
+      },
+    };
+  },
+);
+
+jest.mock(
+  "../../../../App/FeatureSet/Dashboard/src/Components/TelemetryResource/AutoRefreshControl",
+  () => {
+    return {
+      __esModule: true,
+      default: (): null => {
+        return null;
+      },
+    };
+  },
+);
+
+import CloudResourceOverview, {
+  TopInstancesByCpuTitle,
+} from "../../../../App/FeatureSet/Dashboard/src/Pages/Cloud/View/Overview";
+import CloudFleetSummary from "../../../../App/FeatureSet/Dashboard/src/Components/Cloud/CloudFleetSummary";
+import {
+  CloudFleetSummaryTile,
+  summarizeCloudFleet,
+} from "../../../../App/FeatureSet/Dashboard/src/Pages/Cloud/Utils/CloudFleetSummary";
+import { CLOUD_INSTANCE_LIVE_WINDOW_MINUTES } from "../../../../App/FeatureSet/Dashboard/src/Pages/Cloud/Utils/CloudResourceTelemetryScope";
+import {
+  CLOUD_FLEET_METRIC_DESCRIPTIONS,
+  CLOUD_INSTANCE_METRIC_DESCRIPTIONS,
+  CLOUD_METRIC_DESCRIPTIONS,
+  CloudFleetMetric,
+  CloudMetric,
+} from "../../../../App/FeatureSet/Dashboard/src/Components/MetricDescriptions/CloudMetricDescriptions";
+import {
+  expectReadableDescriptionRecord,
+  expectTitleExplained,
+} from "./MetricDescriptionRules";
+import PageComponentProps from "../../../../App/FeatureSet/Dashboard/src/Pages/PageComponentProps";
+import Route from "../../../Types/API/Route";
+import SortOrder from "../../../Types/BaseDatabase/SortOrder";
+
+const PAGE_PROPS: PageComponentProps = {
+  pageRoute: new Route("/cloud/overview"),
+  currentProject: null,
+  hasPaymentMethod: false,
+};
+
+const D: Record<CloudMetric, string> = CLOUD_METRIC_DESCRIPTIONS;
+const F: Record<CloudFleetMetric, string> = CLOUD_FLEET_METRIC_DESCRIPTIONS;
+
+const LIVE_WINDOW: string = `${CLOUD_INSTANCE_LIVE_WINDOW_MINUTES} minutes`;
+
+// Title on the overview -> the text its (i) must show, in page order.
+const TILES: Array<[string, CloudMetric]> = [
+  ["CPU", "cpu"],
+  ["Memory", "memory"],
+  ["Instances", "instances"],
+  ["Requests", "requests"],
+  ["Error rate", "errorRate"],
+  ["p95 latency", "p95Latency"],
+];
+
+const CHARTS: Array<[string, CloudMetric]> = [
+  ["Requests", "requestsChart"],
+  ["Memory", "memoryChart"],
+];
+
+const TOP_INSTANCES: [string, CloudMetric] = [
+  "Top instances by CPU",
+  "topInstancesByCpu",
+];
+
+// The tiles that read each live instance's latest snapshot.
+const SNAPSHOT_KEYS: Array<CloudMetric> = [
+  "cpu",
+  "memory",
+  "instances",
+  "topInstancesByCpu",
+];
+
+// The tiles and charts that follow the time picker.
+const RANGED_KEYS: Array<CloudMetric> = [
+  "requests",
+  "errorRate",
+  "p95Latency",
+  "requestsChart",
+  "memoryChart",
+];
+
+const FLEET_TILES: Array<[string, CloudFleetMetric]> = [
+  ["Environments", "environments"],
+  ["Connected", "connected"],
+  ["Disconnected", "disconnected"],
+  ["Live instances", "liveInstances"],
+];
+
+describe("CLOUD_METRIC_DESCRIPTIONS", () => {
+  test("every text reads as a short, finished, jargon-explaining sentence", () => {
+    expectReadableDescriptionRecord(D, "CLOUD_METRIC_DESCRIPTIONS");
+  });
+
+  test("covers exactly the tiles, charts and cards on the overview", () => {
+    expect(
+      [...TILES, ...CHARTS, TOP_INSTANCES]
+        .map((entry: [string, CloudMetric]): string => {
+          return entry[1];
+        })
+        .sort(),
+    ).toEqual(Object.keys(D).sort());
+  });
+
+  test.each([...TILES, ...CHARTS, TOP_INSTANCES])(
+    "the %s text explains any percentile its title names",
+    (title: string, key: CloudMetric) => {
+      expectTitleExplained(title, D[key]);
+    },
+  );
+
+  test("the two clocks are covered: every text belongs to exactly one", () => {
+    expect([...SNAPSHOT_KEYS, ...RANGED_KEYS].sort()).toEqual(
+      Object.keys(D).sort(),
+    );
+  });
+
+  test.each(SNAPSHOT_KEYS)(
+    "%s says it is a live snapshot over the live window, not the time range",
+    (key: CloudMetric) => {
+      expect(D[key]).toContain(LIVE_WINDOW);
+      expect(D[key]).toMatch(/ignores the selected time range/);
+    },
+  );
+
+  test.each(RANGED_KEYS)(
+    "%s follows the time range and never claims to ignore it",
+    (key: CloudMetric) => {
+      expect(D[key]).toMatch(/selected range|each interval/);
+      expect(D[key]).not.toMatch(/ignores the selected/);
+    },
+  );
+
+  test("CPU is an average of latest readings; Memory is a sum", () => {
+    expect(D.cpu).toMatch(/^Average of the latest CPU/);
+    // The page filters out instances whose latestCpuPercent is not a number.
+    expect(D.cpu).toMatch(/ones with none are left out/);
+    expect(D.memory).toMatch(/added up/);
+    expect(D.memory).toMatch(/latest reading/);
+    // A latest reading can be old; the text must not promise "right now".
+    expect(D.memory).not.toMatch(/right now/);
+  });
+
+  test("CPU says what 100% means and that it is not capped", () => {
+    /*
+     * latestCpuPercent is the receiver's own percentage (per vCPU / core
+     * for ECS task metrics and docker_stats), so a multi-core task reads
+     * above 100%.
+     */
+    // The same reading as the Instances tab's CPU column, so the same words.
+    for (const text of [D.cpu, CLOUD_INSTANCE_METRIC_DESCRIPTIONS.cpu]) {
+      expect(text).toContain(
+        "100% is one full CPU core or all the CPU the task was given, so it can read above 100%",
+      );
+    }
+  });
+
+  test("the p95 tile says each interval counts equally, like the other overview p95 tiles", () => {
+    expect(D.p95Latency).toContain(
+      "then averaged over the selected range, so quiet and busy intervals count equally",
+    );
+  });
+
+  test("Requests is honest that it counts spans across the environment", () => {
+    expect(D.requests).toMatch(/Every span/);
+    expect(D.requests).toMatch(/cloud platform, account and region/);
+    // "usually": a trace with a single span is one request and one span.
+    expect(D.requests).toMatch(/usually higher than the request count/);
+  });
+
+  test("the p95 tile is an average of per-interval p95s", () => {
+    expect(D.p95Latency).toMatch(/95%/);
+    expect(D.p95Latency).toMatch(/for each interval, then averaged/);
+  });
+
+  test("Error rate names the thresholds the tile uses", () => {
+    expect(D.errorRate).toContain("amber at 1%");
+    expect(D.errorRate).toContain("red at 5%");
+  });
+
+  test("the Memory chart warns that repeated readings add up", () => {
+    // The page fetches container.memory.usage with AggregationType.Sum.
+    expect(D.memoryChart).toContain("container.memory.usage");
+    expect(D.memoryChart).toMatch(/counts more than once/);
+    expect(D.memoryChart).toMatch(/trend, not a total/);
+  });
+
+  test("Top instances names its cap and its order", () => {
+    expect(D.topInstancesByCpu).toMatch(/Up to five/);
+    expect(D.topInstancesByCpu).toMatch(/highest first/);
+  });
+
+  test("Top instances is honest that a task with no CPU reading can lead", () => {
+    /*
+     * The list is sorted by latestCpuPercent DESC in Postgres, which puts
+     * NULLs first, and the page keeps the server's order.
+     */
+    expect(D.topInstancesByCpu).toMatch(
+      /not reported CPU show a dash and may be listed first/,
+    );
+  });
+});
+
+describe("CLOUD_FLEET_METRIC_DESCRIPTIONS", () => {
+  test("every text reads as a short, finished, jargon-explaining sentence", () => {
+    expectReadableDescriptionRecord(F, "CLOUD_FLEET_METRIC_DESCRIPTIONS");
+  });
+
+  test("covers exactly the four tiles summarizeCloudFleet returns", () => {
+    const tiles: Array<CloudFleetSummaryTile> = summarizeCloudFleet({
+      total: 0,
+      connected: 0,
+      disconnected: 0,
+      byProvider: {},
+      liveInstances: 0,
+    });
+
+    expect(
+      tiles.map((tile: CloudFleetSummaryTile): [string, string] => {
+        return [tile.title, tile.description];
+      }),
+    ).toEqual(
+      FLEET_TILES.map((entry: [string, CloudFleetMetric]): [string, string] => {
+        return [entry[0], F[entry[1]]];
+      }),
+    );
+  });
+
+  test("the description does not depend on the counts", () => {
+    const empty: Array<CloudFleetSummaryTile> = summarizeCloudFleet({
+      total: 0,
+      connected: 0,
+      disconnected: 0,
+      byProvider: {},
+      liveInstances: 0,
+    });
+    const busy: Array<CloudFleetSummaryTile> = summarizeCloudFleet({
+      total: 9,
+      connected: 2,
+      disconnected: 7,
+      byProvider: { aws: 9 },
+      liveInstances: 40,
+    });
+
+    expect(
+      busy.map((tile: CloudFleetSummaryTile): string => {
+        return tile.description;
+      }),
+    ).toEqual(
+      empty.map((tile: CloudFleetSummaryTile): string => {
+        return tile.description;
+      }),
+    );
+  });
+
+  test("Environments and Live instances say which rows they count", () => {
+    // fetchCloudFleetCounts: environments are isArchived=false; instances are not.
+    expect(F.environments).toMatch(/not counting archived ones/);
+    expect(F.liveInstances).toMatch(/archived ones included/);
+    expect(F.liveInstances).toContain(LIVE_WINDOW);
+  });
+
+  test("Connected and Disconnected describe the status sweep", () => {
+    expect(F.connected).toMatch(/15 to 20 minutes/);
+    expect(F.disconnected).toMatch(/never reported/);
+  });
+});
+
+async function hover(trigger: HTMLElement): Promise<void> {
+  fireEvent.mouseEnter(trigger);
+  await act(async () => {
+    jest.advanceTimersByTime(200);
+  });
+}
+
+function infoButtons(title: string): Array<HTMLElement> {
+  return screen.getAllByRole("button", { name: `About ${title}` });
+}
+
+async function renderOverview(): Promise<void> {
+  render(
+    <MemoryRouter>
+      <CloudResourceOverview {...PAGE_PROPS} />
+    </MemoryRouter>,
+  );
+
+  await screen.findByText("task-a");
+  await screen.findByText("250 ms");
+}
+
+describe("Cloud environment overview page", () => {
+  beforeEach(() => {
+    jest.useFakeTimers();
+    mockAggregateCalls.length = 0;
+    mockListCalls.length = 0;
+    mockInstances = [...MOCK_DEFAULT_INSTANCES];
+    mockHoldAggregates = false;
+    mockHoldInstances = false;
+  });
+
+  afterEach(() => {
+    cleanup();
+    jest.useRealTimers();
+    for (const key of Object.keys(mockTranslations)) {
+      delete mockTranslations[key];
+    }
+  });
+
+  test("renders one (i) per tile, per chart and for the top-instances card", async () => {
+    await renderOverview();
+
+    expect(screen.getAllByRole("button", { name: /^About / })).toHaveLength(
+      TILES.length + CHARTS.length + 1,
+    );
+  });
+
+  test("shows the numbers the descriptions promise", async () => {
+    await renderOverview();
+
+    // Live instances only: (80 + 20) / 2, 512 + 256 MiB, two tasks.
+    expect(screen.getByText("50.0%")).toBeInTheDocument();
+    expect(screen.getByText("768 MiB")).toBeInTheDocument();
+    expect(screen.queryByText("task-gone")).not.toBeInTheDocument();
+  });
+
+  test.each(TILES)(
+    "the %s tile explains itself on hover",
+    async (title: string, key: CloudMetric) => {
+      await renderOverview();
+
+      await hover(infoButtons(title)[0]!);
+
+      expect(screen.getByRole("tooltip")).toHaveTextContent(D[key]);
+    },
+  );
+
+  test.each(CHARTS)(
+    "the %s chart explains itself on hover",
+    async (title: string, key: CloudMetric) => {
+      await renderOverview();
+
+      const buttons: Array<HTMLElement> = infoButtons(title);
+
+      // The tile of the same name renders first.
+      expect(buttons).toHaveLength(2);
+
+      await hover(buttons[1]!);
+
+      expect(screen.getByRole("tooltip")).toHaveTextContent(D[key]);
+    },
+  );
+
+  test("the Top instances by CPU card explains itself from its heading", async () => {
+    await renderOverview();
+
+    const button: HTMLElement = screen.getByRole("button", {
+      name: `About ${TOP_INSTANCES[0]}`,
+    });
+
+    expect(button.closest("h2")).not.toBeNull();
+    expect(button.closest("h2")).toHaveTextContent(TOP_INSTANCES[0]);
+
+    await hover(button);
+
+    expect(screen.getByRole("tooltip")).toHaveTextContent(D.topInstancesByCpu);
+  });
+
+  test("no (i) sits inside a link or another button", async () => {
+    await renderOverview();
+
+    for (const button of screen.getAllByRole("button", { name: /^About / })) {
+      expect(button.closest("a")).toBeNull();
+      expect(button.parentElement?.closest("button") ?? null).toBeNull();
+    }
+  });
+
+  test("the Memory chart it explains really is a sum of container.memory.usage", async () => {
+    await renderOverview();
+
+    const metricCalls: Array<MockAggregateCall> = mockAggregateCalls.filter(
+      (call: MockAggregateCall): boolean => {
+        return call.modelType.name === "Metric";
+      },
+    );
+
+    expect(metricCalls).toHaveLength(1);
+    expect(metricCalls[0]!.aggregateBy.query["name"]).toBe(
+      "container.memory.usage",
+    );
+    expect(metricCalls[0]!.aggregateBy.aggregationType).toBe("Sum");
+  });
+
+  test("CPU leaves out a live task with no CPU reading, as its text says", async () => {
+    mockInstances = [
+      ...MOCK_DEFAULT_INSTANCES,
+      {
+        instanceName: "task-quiet",
+        latestMemoryBytes: 128 * MOCK_MIB,
+        minutesAgo: 3,
+      },
+    ];
+
+    await renderOverview();
+
+    // (80 + 20) / 2 - not (80 + 20 + 0) / 3 = 33.3%.
+    expect(screen.getByText("50.0%")).toBeInTheDocument();
+    expect(screen.queryByText("33.3%")).not.toBeInTheDocument();
+    // Memory still adds the quiet task: 512 + 256 + 128 MiB.
+    expect(screen.getByText("896 MiB")).toBeInTheDocument();
+    expect(D.cpu).toMatch(/ones with none are left out/);
+  });
+
+  test("CPU is not capped at 100%, as its text says", async () => {
+    mockInstances = [
+      {
+        instanceName: "task-a",
+        latestCpuPercent: 250,
+        latestMemoryBytes: 512 * MOCK_MIB,
+        minutesAgo: 1,
+      },
+    ];
+
+    await renderOverview();
+
+    // Tile and top-instances row both show the raw percentage.
+    expect(screen.getAllByText("250.0%").length).toBeGreaterThanOrEqual(2);
+    expect(D.cpu).toContain("can read above 100%");
+  });
+
+  test("Top instances keeps the server's CPU order, so a task with no CPU can lead", async () => {
+    // As Postgres returns ORDER BY latestCpuPercent DESC: NULLs first.
+    mockInstances = [
+      {
+        instanceName: "task-quiet",
+        latestMemoryBytes: 128 * MOCK_MIB,
+        minutesAgo: 3,
+      },
+      ...MOCK_DEFAULT_INSTANCES,
+    ];
+
+    await renderOverview();
+
+    const instanceList: Array<MockListCall> = mockListCalls.filter(
+      (call: MockListCall): boolean => {
+        return call.modelType.name === "CloudResourceInstance";
+      },
+    );
+
+    expect(instanceList).toHaveLength(1);
+    expect(instanceList[0]!.sort).toEqual({
+      latestCpuPercent: SortOrder.Descending,
+    });
+
+    const names: Array<string> = ["task-quiet", "task-a", "task-b"];
+    const rendered: Array<string> = names.filter((name: string): boolean => {
+      return screen.queryByText(name) !== null;
+    });
+
+    expect(rendered).toEqual(names);
+
+    const first: HTMLElement = screen.getByText("task-quiet");
+    const second: HTMLElement = screen.getByText("task-a");
+
+    // task-quiet is rendered before task-a.
+    expect(
+      first.compareDocumentPosition(second) & Node.DOCUMENT_POSITION_FOLLOWING,
+    ).toBeTruthy();
+    // Its CPU cell is a dash, as the text says.
+    expect(
+      within(first.parentElement as HTMLElement).getByText("—"),
+    ).toBeInTheDocument();
+    expect(D.topInstancesByCpu).toMatch(/show a dash and may be listed first/);
+  });
+
+  test("every tile keeps its (i) while its numbers are still loading", async () => {
+    mockHoldAggregates = true;
+    mockHoldInstances = true;
+
+    render(
+      <MemoryRouter>
+        <CloudResourceOverview {...PAGE_PROPS} />
+      </MemoryRouter>,
+    );
+
+    await screen.findByText("Loading CPU");
+
+    for (const [title, key] of TILES) {
+      expect(screen.getByText(`Loading ${title}`)).toBeInTheDocument();
+
+      const button: HTMLElement = infoButtons(title)[0]!;
+
+      await hover(button);
+
+      const tooltips: Array<HTMLElement> = screen.getAllByRole("tooltip");
+
+      expect(tooltips[tooltips.length - 1]).toHaveTextContent(D[key]);
+      fireEvent.mouseLeave(button);
+    }
+
+    // The chart cards are loading too, and explain themselves as well.
+    for (const [title, key] of CHARTS) {
+      const buttons: Array<HTMLElement> = infoButtons(title);
+
+      await hover(buttons[buttons.length - 1]!);
+
+      const tooltips: Array<HTMLElement> = screen.getAllByRole("tooltip");
+
+      expect(tooltips[tooltips.length - 1]).toHaveTextContent(D[key]);
+      fireEvent.mouseLeave(buttons[buttons.length - 1]!);
+    }
+  });
+
+  test("the span tiles it explains are scoped by the cloud attributes", async () => {
+    await renderOverview();
+
+    const spanCalls: Array<MockAggregateCall> = mockAggregateCalls.filter(
+      (call: MockAggregateCall): boolean => {
+        return call.modelType.name === "Span";
+      },
+    );
+
+    expect(spanCalls.length).toBeGreaterThan(0);
+
+    for (const call of spanCalls) {
+      expect(call.aggregateBy.query["attributes"]).toEqual({
+        "resource.cloud.platform": "aws_ecs",
+        "resource.cloud.account.id": "123456789012",
+        "resource.cloud.region": "us-east-1",
+      });
+    }
+  });
+});
+
+describe("TopInstancesByCpuTitle", () => {
+  beforeEach(() => {
+    jest.useFakeTimers();
+  });
+
+  afterEach(() => {
+    cleanup();
+    jest.useRealTimers();
+    for (const key of Object.keys(mockTranslations)) {
+      delete mockTranslations[key];
+    }
+  });
+
+  test("renders the title and an (i) that explains it", async () => {
+    render(<TopInstancesByCpuTitle />);
+
+    expect(screen.getByText("Top instances by CPU")).toBeInTheDocument();
+
+    await hover(
+      screen.getByRole("button", { name: "About Top instances by CPU" }),
+    );
+
+    expect(screen.getByRole("tooltip")).toHaveTextContent(D.topInstancesByCpu);
+  });
+
+  test("keeps the translation a string title would have had", () => {
+    /*
+     * Card translates a string title but passes an element through, so the
+     * element has to translate itself or a localized dashboard loses it.
+     */
+    mockTranslations["Top instances by CPU"] = "Instancias con más CPU";
+
+    render(<TopInstancesByCpuTitle />);
+
+    expect(screen.getByText("Instancias con más CPU")).toBeInTheDocument();
+    expect(
+      screen.getByRole("button", { name: "About Instancias con más CPU" }),
+    ).toBeInTheDocument();
+  });
+});
+
+describe("Cloud fleet summary strip", () => {
+  beforeEach(() => {
+    jest.useFakeTimers();
+  });
+
+  afterEach(() => {
+    cleanup();
+    jest.useRealTimers();
+  });
+
+  async function renderStrip(): Promise<void> {
+    render(<CloudFleetSummary />);
+    await screen.findByTestId("cloud-fleet-summary");
+  }
+
+  test("every tile carries exactly one (i)", async () => {
+    await renderStrip();
+
+    const tiles: Array<HTMLElement> = screen.getAllByTestId(
+      "cloud-fleet-summary-tile",
+    );
+
+    expect(tiles).toHaveLength(FLEET_TILES.length);
+
+    tiles.forEach((tile: HTMLElement, index: number): void => {
+      const title: string = FLEET_TILES[index]![0];
+
+      expect(
+        within(tile).getAllByRole("button", { name: /^About / }),
+      ).toHaveLength(1);
+      expect(
+        within(tile).getByRole("button", { name: `About ${title}` }),
+      ).toBeInTheDocument();
+    });
+  });
+
+  test.each(FLEET_TILES)(
+    "the %s tile explains itself on hover",
+    async (title: string, key: CloudFleetMetric) => {
+      await renderStrip();
+
+      await hover(screen.getByRole("button", { name: `About ${title}` }));
+
+      expect(screen.getByRole("tooltip")).toHaveTextContent(F[key]);
+    },
+  );
+
+  test("the tooltip text is not rendered until someone asks for it", async () => {
+    await renderStrip();
+
+    for (const [, key] of FLEET_TILES) {
+      expect(screen.queryByText(F[key])).not.toBeInTheDocument();
+    }
+  });
+});