@skyhook-io/radar-app 1.13.4 → 1.14.0

package/src/components/diagnose/investigationEvidence/adapters/diagnose.test.ts

@@ -0,0 +1,573 @@
+import { describe, expect, it } from "vitest";
+import { investigationEvidenceSubjectRef } from "../index";
+import {
+  criticalIssue,
+  deployment,
+  groupsOf,
+  project,
+  tool,
+  warningEvent,
+} from "../evidenceFixtures";
+
+describe("semantic diagnose evidence projection", () => {
+  it("promotes only classified smoking guns and audits every partiality signal", () => {
+    const result = project([
+      tool("diagnose-1", "diagnose", {
+        resource: deployment,
+        resourceContext: {
+          tier: "diagnostic",
+          issueSummary: {
+            count: 1,
+            highestSeverity: "critical",
+            topReason: "CrashLoopBackOff",
+          },
+          // Static posture stays attached to Resource; it is never its own Key.
+          auditSummary: {
+            count: 3,
+            highestSeverity: "high",
+            topFinding: "run-as-root",
+          },
+          referencedBy: { total: 8, items: [], truncated: true },
+          omitted: [{ field: "uses.secrets", reason: "rbac_denied" }],
+        },
+        pods: 2,
+        relatedIssues: [criticalIssue],
+        startupBlockers: [
+          {
+            kind: "Pod",
+            name: "api-abc",
+            reason: "ImagePullBackOff",
+            severity: "critical",
+            message: "The image tag does not exist.",
+          },
+        ],
+        crashCause: [
+          {
+            pods: ["api-abc"],
+            container: "api",
+            state: "terminated",
+            reason: "Error",
+            exitCode: 1,
+            logLine: "FATAL: missing DATABASE_URL",
+            logSource: "previous",
+            logLineSelection: "fatal_pattern",
+          },
+        ],
+        crashCauseTruncated: true,
+        logsCurrent: [
+          {
+            pod: "api-abc",
+            container: "api",
+            logs: {
+              lines: ["ERROR database unavailable"],
+              totalLines: 30,
+              matchedLines: 1,
+              fallback: false,
+            },
+          },
+          {
+            pod: "api-def",
+            container: "api",
+            logs: {
+              lines: ["server started"],
+              totalLines: 1,
+              matchedLines: 0,
+              fallback: true,
+            },
+          },
+        ],
+        logCoverage: {
+          resolvedPods: 5,
+          selectedPods: 2,
+          selectionTruncated: true,
+          shownLines: 2,
+          totalLines: 8,
+          contentTruncated: true,
+        },
+        events: [warningEvent],
+        eventsTotalGroups: 3,
+        recentChanges: [
+          {
+            kind: "Deployment",
+            namespace: "shop",
+            name: "api",
+            changeType: "update",
+            timestamp: "2026-09-02T09:55:00Z",
+            summary: "image changed",
+          },
+        ],
+        recentChangesSaturated: true,
+        dnsContext: { signals: ["lookup failures in application logs"] },
+        narrowHint: "log fan-out selected 2 of 5 pods",
+        warnings: ["Managed by Helm; edit the source of truth."],
+      }),
+    ]);
+
+    expect(groupsOf(result.groups, "issue")[0].latest).toMatchObject({
+      tier: "key",
+      relevance: "target",
+    });
+    expect(groupsOf(result.groups, "startup")[0].latest.tier).toBe("key");
+    expect(groupsOf(result.groups, "crash")[0].latest.tier).toBe("key");
+    expect(groupsOf(result.groups, "startup")[0].latest.relevance).toBe(
+      "producer-related",
+    );
+    expect(groupsOf(result.groups, "startup")[0].latest.data).toMatchObject({
+      type: "startup",
+      subject: { kind: "Pod", namespace: "shop", name: "api-abc" },
+    });
+    expect(groupsOf(result.groups, "crash")[0].latest.relevance).toBe(
+      "producer-related",
+    );
+    // The detailed critical Issue owns the Key slot; its aggregate resource
+    // rollup stays supporting rather than counting the same signal twice.
+    expect(groupsOf(result.groups, "resource")[0].latest.tier).toBe(
+      "supporting",
+    );
+    expect(groupsOf(result.groups, "resource")[0].latest.data).toMatchObject({
+      type: "resource",
+      warnings: ["Managed by Helm; edit the source of truth."],
+    });
+    expect(
+      groupsOf(result.groups, "logs").map((group) => group.latest.tier),
+    ).toEqual(["supporting", "context"]);
+    expect(groupsOf(result.groups, "dns")).toHaveLength(1);
+    expect(
+      result.groups.some((group) => group.identity.includes("audit")),
+    ).toBe(false);
+
+    const limitationText = result.limitations
+      .map((item) => item.message)
+      .join("\n");
+    expect(limitationText).toContain("selected 2 of 5");
+    expect(limitationText).toContain("includes 2 of 8 selected log lines");
+    expect(limitationText).toContain("not the container's full log history");
+    expect(limitationText).toContain("part of the crash-cause candidates");
+    expect(limitationText).toContain("received 1 of 3 event groups");
+    expect(limitationText).toContain("rbac denied");
+    expect(limitationText).toContain("referenced-by relationships");
+    expect(limitationText).toContain("recent-change result limit");
+    expect(result.coverage).toEqual({
+      attempted: 1,
+      projected: 1,
+      limited: 1,
+      checked: 0,
+    });
+    expect(result.sources[0].primaryGroupId).toBe(
+      groupsOf(result.groups, "issue")[0].id,
+    );
+  });
+
+  it("uses a critical live issueSummary as Key when no detailed issue duplicates it", () => {
+    const result = project([
+      tool("resource-summary", "diagnose", {
+        resource: deployment,
+        resourceContext: {
+          tier: "basic",
+          issueSummary: {
+            count: 2,
+            highestSeverity: "critical",
+            topReason: "Workload unavailable",
+          },
+        },
+        pods: 0,
+        events: [],
+        recentChanges: [],
+      }),
+    ]);
+    expect(groupsOf(result.groups, "resource")[0].latest.tier).toBe("key");
+  });
+
+  it("keeps an explicit replica shortfall or failing controller condition visible as Supporting", () => {
+    const result = project([
+      tool("diagnose-state", "diagnose", {
+        resource: deployment,
+        resourceContext: {
+          tier: "basic",
+          workloadSummary: {
+            replicas: { desired: 2, ready: 1, available: 1, unavailable: 1 },
+          },
+          statusSummary: {
+            conditions: [
+              {
+                type: "ReplicaFailure",
+                status: "True",
+                reason: "FailedCreate",
+              },
+            ],
+          },
+        },
+        pods: 1,
+        events: [],
+        recentChanges: [],
+      }),
+    ]);
+    expect(groupsOf(result.groups, "resource")[0].latest).toMatchObject({
+      tier: "supporting",
+      tone: "warning",
+    });
+  });
+});
+
+describe("diagnose metrics evidence", () => {
+  const window = {
+    start: "2026-09-06T07:00:00Z",
+    end: "2026-09-06T08:00:00Z",
+    step: "1m2s",
+  };
+  const samples = [
+    { timestamp: Date.parse(window.start) / 1000, value: 0.003457 },
+    { timestamp: Date.parse(window.end) / 1000, value: 0.004 },
+  ];
+  function vitals(patch: Record<string, unknown> = {}) {
+    return {
+      window,
+      pods: 2,
+      series: [
+        {
+          category: "cpu",
+          unit: "cores",
+          query:
+            "sum(rate(container_cpu_usage_seconds_total{container!='',namespace='shop',pod=~'^(api-a|api-b)$'}[5m]))",
+          series: [{ labels: {}, dataPoints: samples }],
+        },
+        {
+          category: "memory",
+          unit: "bytes",
+          query:
+            "sum(max by (pod,namespace,container) (container_memory_working_set_bytes{container!='',namespace='shop',pod=~'^(api-a|api-b)$'}))",
+          series: [{ labels: {}, dataPoints: samples }],
+        },
+        {
+          category: "restarts",
+          unit: "count",
+          query:
+            "sum(round(increase(kube_pod_container_status_restarts_total{namespace='shop',pod=~'^(api-a|api-b)$'}[1h])))",
+          series: [],
+        },
+      ],
+      ...patch,
+    };
+  }
+  const args = { kind: "Deployment", namespace: "shop", name: "api" };
+
+  it("charts each captured category as target evidence with the diagnosed resource as subject", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        { resource: deployment, pods: 2, metrics: vitals() },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    const groups = groupsOf(projection.groups, "metrics");
+    expect(groups.map((group) => group.latest.title)).toEqual([
+      "CPU usage · Deployment shop/api",
+      "Memory working set · Deployment shop/api",
+      "Restarts · Deployment shop/api",
+    ]);
+    for (const group of groups) {
+      expect(group.latest.relevance).toBe("target");
+      expect(group.latest.tier).toBe("supporting");
+      expect(group.latest.tone).toBe("neutral");
+      const data = group.latest.data;
+      if (data.type !== "metrics") throw new Error("expected metrics");
+      expect(data.origin).toBe("diagnose");
+      expect(data.mode).toBe("range");
+      expect(data.start).toBe(window.start);
+      expect(data.end).toBe(window.end);
+      expect(data.step).toBe(window.step);
+      expect(data.pods).toBe(2);
+      expect(data.partial).toBe(false);
+      expect(data.truncated).toBe(false);
+      expect(data.subject).toEqual({
+        kind: "Deployment",
+        group: "apps",
+        namespace: "shop",
+        name: "api",
+      });
+      expect(investigationEvidenceSubjectRef(data)).toEqual(data.subject);
+    }
+    const [cpu, , restarts] = groups;
+    expect(cpu.latest.summary).toBe("2 current pods · 60m window · 1m2s step");
+    const cpuData = cpu.latest.data;
+    if (cpuData.type !== "metrics") throw new Error("expected metrics");
+    expect(cpuData.unit).toBe("cores");
+    expect(cpuData.label).toBe("CPU usage");
+    expect(cpuData.query).toContain("pod=~'^(api-a|api-b)$'");
+    expect(cpuData.series).toEqual([{ labels: {}, dataPoints: samples }]);
+    expect(restarts.latest.summary).toBe(
+      "No samples in the window · 60m window · 1m2s step",
+    );
+    expect(
+      projection.limitations.filter(
+        (item) => item.source === "Workload metrics",
+      ),
+    ).toHaveLength(0);
+  });
+
+  it("merges a re-diagnose of the same workload into the same chart per category", () => {
+    const later = vitals({
+      window: { ...window, end: "2026-09-06T08:30:00Z" },
+    });
+    const projection = project(
+      [
+        tool(
+          "diag-1",
+          "diagnose",
+          { resource: deployment, pods: 2, metrics: vitals() },
+          { summary: JSON.stringify(args) },
+        ),
+      ],
+      [
+        tool(
+          "diag-2",
+          "diagnose",
+          { resource: deployment, pods: 2, metrics: later },
+          { summary: JSON.stringify(args) },
+        ),
+      ],
+    );
+    const groups = groupsOf(projection.groups, "metrics");
+    expect(groups).toHaveLength(3);
+    expect(groups.every((group) => group.observations.length === 2)).toBe(true);
+    const data = groups[0].latest.data;
+    if (data.type !== "metrics") throw new Error("expected metrics");
+    expect(data.end).toBe("2026-09-06T08:30:00Z");
+  });
+
+  it("reports a producer error as a limitation and still charts what answered", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        {
+          resource: deployment,
+          pods: 2,
+          metrics: vitals({
+            series: [vitals().series[1]],
+            error:
+              "cpu: prom: query error from prometheus: cpu exploded (execution)",
+          }),
+        },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    expect(groupsOf(projection.groups, "metrics")).toHaveLength(1);
+    const limitation = projection.limitations.find(
+      (item) => item.source === "Workload metrics",
+    );
+    expect(limitation?.kind).toBe("error");
+    expect(limitation?.message).toContain("cpu exploded");
+  });
+
+  it("reports an unreachable Prometheus without inventing a chart", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        {
+          resource: deployment,
+          pods: 2,
+          metrics: vitals({
+            series: [],
+            error:
+              "prometheus unreachable: dial tcp 10.0.0.9:9090: connection refused",
+          }),
+        },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    expect(groupsOf(projection.groups, "metrics")).toHaveLength(0);
+    const limitation = projection.limitations.find(
+      (item) => item.source === "Workload metrics",
+    );
+    expect(limitation?.kind).toBe("error");
+    expect(limitation?.message).toBe(
+      "prometheus unreachable: dial tcp 10.0.0.9:9090: connection refused",
+    );
+  });
+
+  it("says nothing when diagnose carried no metrics field", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        { resource: deployment, pods: 2 },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    expect(groupsOf(projection.groups, "metrics")).toHaveLength(0);
+    expect(
+      projection.limitations.some((item) => item.source === "Workload metrics"),
+    ).toBe(false);
+  });
+
+  it("notes a partial pod set in the summary and keeps the cap on the card", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        {
+          resource: deployment,
+          pods: 55,
+          metrics: vitals({ pods: 50, partial: true, omittedPods: 5 }),
+        },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    const [cpu] = groupsOf(projection.groups, "metrics");
+    expect(cpu.latest.summary).toBe(
+      "first 50 of 55 current pods · 60m window · 1m2s step · partial pod set",
+    );
+    const data = cpu.latest.data;
+    if (data.type !== "metrics") throw new Error("expected metrics");
+    expect(data.pods).toBe(50);
+    expect(data.partial).toBe(true);
+  });
+
+  it("keeps a neighbour's vitals broader with the neighbour as subject", () => {
+    const worker = {
+      ...deployment,
+      metadata: { namespace: "shop", name: "worker" },
+    };
+    const projection = project([
+      tool(
+        "diag-worker",
+        "diagnose",
+        { resource: worker, pods: 2, metrics: vitals() },
+        {
+          summary: JSON.stringify({
+            kind: "Deployment",
+            namespace: "shop",
+            name: "worker",
+          }),
+        },
+      ),
+    ]);
+    const groups = groupsOf(projection.groups, "metrics");
+    expect(groups).toHaveLength(3);
+    for (const group of groups) {
+      expect(group.latest.relevance).toBe("broader");
+      expect(group.latest.tier).toBe("context");
+      const data = group.latest.data;
+      if (data.type !== "metrics") throw new Error("expected metrics");
+      expect(data.subject).toEqual({
+        kind: "Deployment",
+        group: "apps",
+        namespace: "shop",
+        name: "worker",
+      });
+    }
+  });
+
+  it("rejects a malformed metrics field instead of charting it", () => {
+    const malformed: Record<string, unknown>[] = [
+      { window: { start: window.start }, pods: "two", series: [] },
+      { window: { ...window, end: "not a date" }, pods: 2, series: [] },
+      { window: { ...window, end: window.start }, pods: 2, series: [] },
+      { window, pods: -1, series: [] },
+      { window, pods: 1.5, series: [] },
+      { window, pods: 2, omittedPods: -3, series: [] },
+    ];
+    for (const metrics of malformed) {
+      const projection = project([
+        tool(
+          "diag",
+          "diagnose",
+          { resource: deployment, pods: 2, metrics },
+          { summary: JSON.stringify(args) },
+        ),
+      ]);
+      expect(groupsOf(projection.groups, "metrics")).toHaveLength(0);
+      expect(
+        projection.limitations.some(
+          (item) =>
+            item.source === "Workload metrics" && item.kind === "unknown",
+        ),
+      ).toBe(true);
+    }
+  });
+
+  it("keeps same-named workloads in different API groups on separate charts", () => {
+    const rollout = {
+      ...deployment,
+      apiVersion: "argoproj.io/v1alpha1",
+      kind: "Rollout",
+    };
+    const projection = project([
+      tool(
+        "diag-deployment",
+        "diagnose",
+        { resource: deployment, pods: 2, metrics: vitals() },
+        { summary: JSON.stringify(args) },
+      ),
+      tool(
+        "diag-rollout",
+        "diagnose",
+        { resource: rollout, pods: 2, metrics: vitals() },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    const groups = groupsOf(projection.groups, "metrics");
+    expect(groups).toHaveLength(6);
+    expect(groups.every((group) => group.observations.length === 1)).toBe(true);
+    expect(new Set(groups.map((group) => group.identity)).size).toBe(6);
+  });
+
+  it("rejects inherited object keys posing as categories", () => {
+    for (const category of ["__proto__", "constructor", "toString"]) {
+      const projection = project([
+        tool(
+          "diag",
+          "diagnose",
+          {
+            resource: deployment,
+            pods: 2,
+            metrics: vitals({
+              series: [{ ...vitals().series[0], category }],
+            }),
+          },
+          { summary: JSON.stringify(args) },
+        ),
+      ]);
+      expect(groupsOf(projection.groups, "metrics")).toHaveLength(0);
+      expect(
+        projection.limitations.some(
+          (item) =>
+            item.source === "Workload metrics" && item.kind === "unknown",
+        ),
+      ).toBe(true);
+    }
+  });
+
+  it("rejects a category outside the contract or a series without a unit", () => {
+    const projection = project([
+      tool(
+        "diag",
+        "diagnose",
+        {
+          resource: deployment,
+          pods: 2,
+          metrics: vitals({
+            series: [
+              { ...vitals().series[0], category: "latency" },
+              { ...vitals().series[1], unit: undefined },
+              vitals().series[2],
+            ],
+          }),
+        },
+        { summary: JSON.stringify(args) },
+      ),
+    ]);
+    const groups = groupsOf(projection.groups, "metrics");
+    expect(groups.map((group) => group.latest.title)).toEqual([
+      "Restarts · Deployment shop/api",
+    ]);
+    expect(
+      projection.limitations.some(
+        (item) => item.source === "Workload metrics" && item.kind === "unknown",
+      ),
+    ).toBe(true);
+  });
+});