Files
streamline/control-server/test/batch_monitor.test.ts
T
2026-11-16 17:26:44 -08:00

151 lines
7.7 KiB
TypeScript
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
import { test, expect, describe } from "bun:test";
import { readFileSync } from "node:fs";
import path from "node:path";
import { classifyOutcome, INFRA_FAILURE_PREFIXES, phaseOf } from "@bbbug/shared/batch_monitor";
import { buildBatchReport, buildBatchList } from "../src/batch_monitor";
import { AUTH_FAILURE_HINT } from "../../satellite/src/run_outcome";
import type { Task } from "../src/types";
const REPO = path.resolve(import.meta.dir, "../..");
const satText = ["satellite/src/runner.ts", "satellite/src/evaluator.ts"]
.map((f) => readFileSync(path.join(REPO, f), "utf8")).join("\n") + "\n" + AUTH_FAILURE_HINT;
const H = 3_600_000;
const NOW = 1_800_000_000_000;
let seq = 0;
function task(p: Partial<Task>): Task {
seq++;
return {
id: `t${seq}`, title: `bug ${seq}`, meegoId: String(7000000000 + seq), meegoUrl: null,
platform: "ios", status: "awaiting_fix", failureReason: null, executorNodeId: "ios-1",
attempts: 1, tags: ["batch_a"], clusters: ["evaluator"], createdAt: NOW - 10 * H,
statusChangedAt: NOW - H, unattended: true, mrUrl: null,
...p,
} as Task;
}
const node = (id: string, platform: "ios" | "android", extra: Record<string, unknown> = {}) => ({
id, platform, clusters: ["evaluator"], live: true, health: { alerts: [] },
evalInFlight: 0, evalCapacity: 1, fixInFlight: 1, fixCapacity: 1, ...extra,
});
describe("classifyOutcome", () => {
test("fixed / unrepro / archived / cancelled", () => {
expect(classifyOutcome({ status: "manual_review", failureReason: null }).outcome).toBe("fixed");
expect(classifyOutcome({ status: "fix_succeeded", failureReason: null }).outcome).toBe("fixed");
expect(classifyOutcome({ status: "unreproducible", failureReason: null }).outcome).toBe("unrepro");
expect(classifyOutcome({ status: "archived", failureReason: null }).outcome).toBe("task_failed");
expect(classifyOutcome({ status: "cancelled", failureReason: null }).outcome).toBe("aborted");
});
test("a fix whose MR failed stays fixed but says so", () => {
const v = classifyOutcome({ status: "manual_review", failureReason: "MR 提交失败(as x):create_mr: Invalid token" });
expect(v.outcome).toBe("fixed");
expect(v.basis).toContain("MR 提交失败");
});
test("fix_failed splits infra / aborted / task verdict", () => {
expect(classifyOutcome({ status: "fix_failed", failureReason: "API 错误,多次自动重试仍失败(非修复失败,可稍后重排)" }).outcome).toBe("infra");
expect(classifyOutcome({ status: "fix_failed", failureReason: "scheduler caught: boom" }).outcome).toBe("infra");
expect(classifyOutcome({ status: "fix_failed", failureReason: "用户中止" }).outcome).toBe("aborted");
expect(classifyOutcome({ status: "fix_failed", failureReason: "Agent exited with STATUS: FIX_FAILED (or no STATUS line)" }).outcome).toBe("task_failed");
expect(classifyOutcome({ status: "fix_failed", failureReason: null }).outcome).toBe("task_failed");
});
test("info_required: infra vs a real needs-info verdict", () => {
expect(classifyOutcome({ status: "info_required", failureReason: AUTH_FAILURE_HINT }).outcome).toBe("infra");
expect(classifyOutcome({ status: "info_required", failureReason: "工单缺少复现路径" }).outcome).toBe("needs_info");
});
test("every infra prefix still appears in the satellite's wording", () => {
for (const p of INFRA_FAILURE_PREFIXES) expect(satText).toContain(p);
});
test("phases", () => {
expect(phaseOf("initializing")).toBe("evaluating");
expect(phaseOf("pending_confirmation")).toBe("waiting_human");
expect(phaseOf("awaiting_fix")).toBe("queued");
expect(phaseOf("fixing")).toBe("running");
expect(phaseOf("fix_failed")).toBe("settled");
});
});
describe("buildBatchReport", () => {
const tasks = [
task({ status: "manual_review", statusChangedAt: NOW - 2 * H }),
task({ status: "fix_failed", failureReason: "Agent exited with STATUS: FIX_FAILED (or no STATUS line)" }),
task({ status: "fix_failed", failureReason: "API 错误,多次自动重试仍失败(非修复失败,可稍后重排)" }),
task({ status: "fixing", statusChangedAt: NOW - 6 * H }),
task({ status: "awaiting_fix" }),
task({ status: "awaiting_fix", platform: "android", executorNodeId: "and-1" }),
task({ status: "initializing", platform: "android", executorNodeId: null }),
task({ status: "manual_review", tags: ["other"] }),
];
// 本批第 1 条:评估 10 分钟、排队 30 分钟、执行 60 分钟。
const tr = [
{ targetId: "t1", toStatus: "initializing", nodeId: "ios-1", ts: NOW - 5 * H },
{ targetId: "t1", toStatus: "awaiting_fix", nodeId: "ios-1", ts: NOW - 5 * H + 10 * 60_000 },
{ targetId: "t1", toStatus: "fixing", nodeId: "ios-1", ts: NOW - 5 * H + 40 * 60_000 },
{ targetId: "t1", toStatus: "manual_review", nodeId: "ios-1", ts: NOW - 5 * H + 100 * 60_000 },
];
const r = buildBatchReport("batch_a", tasks, tr, [node("ios-1", "ios"), node("and-1", "android", { live: false })], NOW);
test("scopes to the tag and counts outcomes", () => {
expect(r.total).toBe(7);
expect(r.outcomes.fixed).toBe(1);
expect(r.outcomes.task_failed).toBe(1);
expect(r.outcomes.infra).toBe(1);
expect(r.settled).toBe(3);
// 有效成功率不计基础设施失败:1 / (3 − 1)
expect(r.effectiveSuccessRate).toBeCloseTo(0.5);
expect(r.infraRate).toBeCloseTo(1 / 3);
});
test("durations from transitions", () => {
expect(r.durations.eval.p50Min).toBe(10);
expect(r.durations.queue.p50Min).toBe(30);
expect(r.durations.exec.p50Min).toBe(60);
});
test("eta = remaining × median exec ÷ live fix slots; no live slot → null", () => {
const ios = r.platforms.find((p) => p.platform === "ios")!;
expect(ios.remaining).toBe(2);
expect(ios.fixSlots).toBe(1);
expect(ios.etaHours).toBe(2);
const and = r.platforms.find((p) => p.platform === "android")!;
expect(and.fixSlots).toBe(0);
expect(and.etaHours).toBeNull();
});
test("issues: offline node, unroutable, stuck exec, infra failure", () => {
const kinds = r.issues.map((i) => i.kind);
expect(kinds).toContain("node_offline");
expect(kinds).toContain("unroutable"); // android 评估中、无主、唯一的 android 节点离线
expect(kinds).toContain("stuck_exec"); // 6h > max(4h, 3×60min)
expect(kinds).toContain("infra_failed");
expect(r.issues[0].severity).toBe("critical");
});
test("node rows carry batch-local settled / infra counts", () => {
const ios = r.nodes.find((n) => n.id === "ios-1")!;
expect(ios.settled).toBe(3);
expect(ios.infraFailed).toBe(1);
});
test("batch list groups by tag", () => {
const list = buildBatchList(tasks);
expect(list.find((b) => b.tag === "batch_a")).toMatchObject({ total: 7, settled: 3, fixed: 1 });
expect(list.find((b) => b.tag === "other")?.total).toBe(1);
});
});
describe("batch start boundary", () => {
test("phases before the batch started don't count; shown dwell starts at the batch; stuck alerts keep the real age", () => {
const start = NOW - 24 * H;
const old = task({ status: "fixing", statusChangedAt: NOW - 30 * 24 * H, createdAt: NOW - 60 * 24 * H });
const tr = [
{ targetId: old.id, toStatus: "fixing", nodeId: "ios-1", ts: NOW - 40 * 24 * H },
{ targetId: old.id, toStatus: "fix_failed", nodeId: "ios-1", ts: NOW - 39 * 24 * H },
{ targetId: old.id, toStatus: "fixing", nodeId: "ios-1", ts: NOW - 30 * 24 * H },
];
const r = buildBatchReport("batch_a", [old], tr, [node("ios-1", "ios")], NOW, start);
expect(r.durations.exec.n).toBe(0);
expect(r.tasks[0].since).toBe(start);
expect(r.issues.some((i) => i.kind === "stuck_exec" && i.message.includes("720.0"))).toBe(true);
});
});