kimi-code/packages/kap-server/test/instanceRegistry.test.ts
Haozhe 2d874fbd73
fix(agent-core-v2): surface provider auth errors and align agent copy with v1 (#1631)
* fix(agent-core-v2): surface provider rejection on post-refresh 401

- map a 401 that survives forced token refresh to PROVIDER_AUTH_ERROR
  carrying the provider's sanitized message, instead of a misleading
  AUTH_LOGIN_REQUIRED re-login prompt
- propagate provider auth errors through full compaction instead of
  wrapping them as compaction failure
- export sanitizeStatusErrorMessage for reuse

* fix(agent-core-v2): align repeat-reminder and AskUserQuestion copy with v1

- rewrite repeated tool call reminders to redirect instead of prohibit (port of v1 #1518)
- simplify makeReminderText2 to (repeatCount), no longer echoing tool name and args
- treat dismissed AskUserQuestion as no answer, not the recommended pick (port of v1 #1550)
- update toolDedupe test assertions and regenerate affected wire snapshots

* chore: add changeset for v2 reminder and question copy alignment

* fix(kap-server): drain in-flight heartbeat writes before unlinking the instance file

- count writes that passed the released check and await them in release()
  so their atomic rename cannot recreate the file after unlink
- harden the heartbeat test: 1ms cadence plus a post-release re-check
  catches the recreate-after-unlink race (verified failing on old code)

* docs(agent-core-v2): move toProviderAuthError rationale to the file header

Address Codex review: agent-core-v2 keeps comments in the top-of-file
header only, never beside functions
2026-07-13 23:52:22 +08:00

260 lines
9.4 KiB
TypeScript

/**
* Server instance registry semantics (plan P0 — multi-server shared homedir).
*
* Hermetic strategy: every test uses a tmpdir instances dir so the real
* `~/.kimi-code/server/instances` is never touched. Dead-pid simulation uses
* `0x7fffffff` (guaranteed ESRCH on Linux/macOS), mirroring lock.test.ts.
*/
import { existsSync, mkdirSync, mkdtempSync, readFileSync, rmSync, writeFileSync } from 'node:fs';
import { tmpdir } from 'node:os';
import { join } from 'node:path';
import { afterEach, beforeEach, describe, expect, it } from 'vitest';
import {
createInstanceRegistry,
getLiveServerInstance,
listLiveServerInstances,
type ServerInstanceInfo,
} from '../src/instanceRegistry';
let tmpDir: string;
let instancesDir: string;
/** Max signed-32 pid; the kernel never allocates it, so `kill(pid, 0)` → ESRCH. */
const DEAD_PID = 0x7fffffff;
beforeEach(() => {
tmpDir = mkdtempSync(join(tmpdir(), 'kimi-instance-registry-test-'));
instancesDir = join(tmpDir, 'instances');
});
afterEach(() => {
rmSync(tmpDir, { recursive: true, force: true });
});
interface DiskInstance {
server_id: string;
pid: number;
host: string;
port: number;
started_at: number;
heartbeat_at: number;
host_version?: string;
}
function writeInstance(serverId: string, fields: Partial<DiskInstance> & { pid: number }): void {
mkdirSync(instancesDir, { recursive: true });
const disk: DiskInstance = {
server_id: serverId,
pid: fields.pid,
host: fields.host ?? '127.0.0.1',
port: fields.port ?? 58627,
started_at: fields.started_at ?? 1000,
heartbeat_at: fields.heartbeat_at ?? 1000,
...(fields.host_version !== undefined ? { host_version: fields.host_version } : {}),
};
writeFileSync(join(instancesDir, `${serverId}.json`), JSON.stringify(disk));
}
function readInstance(serverId: string): DiskInstance {
return JSON.parse(readFileSync(join(instancesDir, `${serverId}.json`), 'utf8')) as DiskInstance;
}
function sleep(ms: number): Promise<void> {
return new Promise((resolvePromise) => setTimeout(resolvePromise, ms));
}
const baseInfo = {
pid: process.pid,
host: '127.0.0.1',
port: 58627,
startedAt: 1000,
};
describe('createInstanceRegistry — register / release', () => {
it('writes a <serverId>.json file and release removes it', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 2000 });
const reg = await registry.register(baseInfo);
expect(typeof reg.serverId).toBe('string');
expect(reg.serverId.length).toBeGreaterThan(0);
const filePath = join(instancesDir, `${reg.serverId}.json`);
expect(existsSync(filePath)).toBe(true);
expect(readInstance(reg.serverId)).toEqual({
server_id: reg.serverId,
pid: process.pid,
host: '127.0.0.1',
port: 58627,
started_at: 1000,
heartbeat_at: 2000,
});
await reg.release();
expect(existsSync(filePath)).toBe(false);
});
it('records host_version when provided', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
const reg = await registry.register({ ...baseInfo, hostVersion: '1.2.3' });
expect(readInstance(reg.serverId).host_version).toBe('1.2.3');
await reg.release();
});
it('assigns distinct serverIds to concurrent registrations', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
const a = await registry.register(baseInfo);
const b = await registry.register(baseInfo);
expect(a.serverId).not.toBe(b.serverId);
await a.release();
await b.release();
});
it('release is idempotent', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
const reg = await registry.register(baseInfo);
await reg.release();
await expect(reg.release()).resolves.toBeUndefined();
});
it('update is a no-op after release', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
const reg = await registry.register(baseInfo);
const filePath = join(instancesDir, `${reg.serverId}.json`);
await reg.release();
await expect(reg.update({ port: 9999 })).resolves.toBeUndefined();
expect(existsSync(filePath)).toBe(false);
});
});
describe('createInstanceRegistry — stale sweep on register', () => {
it('removes dead-pid entries and keeps live ones when registering', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
// Pre-seed a stale (dead pid) and a live entry before registering.
writeInstance('stale', { pid: DEAD_PID });
writeInstance('live-peer', { pid: process.pid, started_at: 500 });
const reg = await registry.register(baseInfo);
expect(existsSync(join(instancesDir, 'stale.json'))).toBe(false);
expect(existsSync(join(instancesDir, 'live-peer.json'))).toBe(true);
expect(existsSync(join(instancesDir, `${reg.serverId}.json`))).toBe(true);
await reg.release();
});
it('leaves unparseable entries alone (may be a live peer mid-write)', async () => {
const registry = createInstanceRegistry({ instancesDir, now: () => 1 });
mkdirSync(instancesDir, { recursive: true });
writeFileSync(join(instancesDir, 'garbage.json'), '{not valid');
const reg = await registry.register(baseInfo);
expect(existsSync(join(instancesDir, 'garbage.json'))).toBe(true);
await reg.release();
});
});
describe('createInstanceRegistry — listLive', () => {
it('returns live instances, drops dead ones, and sorts by startedAt', async () => {
const registry = createInstanceRegistry({ instancesDir });
writeInstance('dead', { pid: DEAD_PID, started_at: 1 });
writeInstance('older', { pid: process.pid, started_at: 100 });
writeInstance('newer', { pid: process.pid, started_at: 200 });
const live = await registry.listLive();
expect(live.map((i) => i.serverId)).toEqual(['older', 'newer']);
// Dead entry lazily removed as a side effect.
expect(existsSync(join(instancesDir, 'dead.json'))).toBe(false);
});
it('returns an empty array when the directory is missing', async () => {
const registry = createInstanceRegistry({ instancesDir });
await expect(registry.listLive()).resolves.toEqual([]);
});
});
describe('createInstanceRegistry — update', () => {
it('rewrites the port and refreshes heartbeatAt', async () => {
let t = 1000;
const registry = createInstanceRegistry({ instancesDir, now: () => t });
const reg = await registry.register(baseInfo);
expect(readInstance(reg.serverId).port).toBe(58627);
expect(readInstance(reg.serverId).heartbeat_at).toBe(1000);
t = 2000;
await reg.update({ port: 58628 });
const after = readInstance(reg.serverId);
expect(after.port).toBe(58628);
expect(after.heartbeat_at).toBe(2000);
// Other fields preserved.
expect(after.pid).toBe(process.pid);
expect(after.started_at).toBe(1000);
await reg.release();
});
it('refreshes heartbeat without changing port when patch is empty', async () => {
let t = 1000;
const registry = createInstanceRegistry({ instancesDir, now: () => t });
const reg = await registry.register(baseInfo);
t = 3000;
await reg.update({});
const after = readInstance(reg.serverId);
expect(after.port).toBe(58627);
expect(after.heartbeat_at).toBe(3000);
await reg.release();
});
});
describe('createInstanceRegistry — heartbeat', () => {
it('periodically rewrites heartbeatAt until released', async () => {
let tick = 0;
const registry = createInstanceRegistry({
instancesDir,
// 1ms cadence keeps a write in flight at almost every moment, so
// `release()` is exercised against the recreate-after-unlink race.
heartbeatIntervalMs: 1,
now: () => ++tick,
});
const reg = await registry.register(baseInfo);
const first = readInstance(reg.serverId).heartbeat_at;
await sleep(90);
const later = readInstance(reg.serverId).heartbeat_at;
expect(later).toBeGreaterThan(first);
await reg.release();
// File is gone after release, and a heartbeat write that was in flight
// when release() ran must not recreate it.
expect(existsSync(join(instancesDir, `${reg.serverId}.json`))).toBe(false);
await sleep(30);
expect(existsSync(join(instancesDir, `${reg.serverId}.json`))).toBe(false);
});
});
describe('convenience readers', () => {
it('listLiveServerInstances reads <homeDir>/server/instances', async () => {
const dir = join(tmpDir, 'server', 'instances');
const registry = createInstanceRegistry({ instancesDir: dir });
const reg = await registry.register({ ...baseInfo, startedAt: 100 });
const live = await listLiveServerInstances(tmpDir);
expect(live.map((i: ServerInstanceInfo) => i.serverId)).toEqual([reg.serverId]);
await reg.release();
});
it('getLiveServerInstance returns the longest-running live instance', async () => {
const dir = join(tmpDir, 'server', 'instances');
const registry = createInstanceRegistry({ instancesDir: dir });
const older = await registry.register({ ...baseInfo, startedAt: 100 });
const newer = await registry.register({ ...baseInfo, startedAt: 200 });
const first = await getLiveServerInstance(tmpDir);
expect(first?.serverId).toBe(older.serverId);
await older.release();
await newer.release();
});
it('getLiveServerInstance returns undefined when no live instance exists', async () => {
await expect(getLiveServerInstance(tmpDir)).resolves.toBeUndefined();
});
});