fix: retry transient Weixin startup failures

src/channels/weixin/weixin-runtime.mjs

@@ -1,6 +1,26 @@
 import { WeixinApiError } from './weixin-api.mjs';
 import { createWeixinBridgeStatus, WeixinHarnessBridge } from './weixin-bridge.mjs';
 
+const DEFAULT_START_RETRY_DELAYS_MS = Object.freeze([250, 1_000, 3_000]);
+
+function startRetryDelays(value) {
+  if (value === undefined) return [...DEFAULT_START_RETRY_DELAYS_MS];
+  if (!Array.isArray(value)) throw new TypeError('startRetryDelaysMs must be an array');
+  return value.map((wait) => {
+    if (!Number.isFinite(wait) || wait < 0) {
+      throw new TypeError('startRetryDelaysMs must contain non-negative delays');
+    }
+    return wait;
+  });
+}
+
+function retryableStartError(error) {
+  if (!(error instanceof WeixinApiError)) return false;
+  if (error.code === 'network-error' || error.code === 'timeout') return true;
+  return error.code === 'http-error'
+    && (error.status === 408 || error.status === 425 || error.status === 429 || error.status >= 500);
+}
+
 function delay(ms, signal) {
   return new Promise((resolve, reject) => {
     if (signal?.aborted) {
@@ -42,6 +62,7 @@ export class WeixinRuntime {
   #logger;
   #replyTimeoutMs;
   #maxMessageChars;
+  #startRetryDelaysMs;
   #status = createWeixinRuntimeStatus();
   #bridge = null;
   #abortController = null;
@@ -57,6 +78,7 @@ export class WeixinRuntime {
     logger = console,
     replyTimeoutMs = 600_000,
     maxMessageChars = 4_000,
+    startRetryDelaysMs,
   }) {
     if (!api || !config || !token || !harness || !state) {
       throw new TypeError('WeixinRuntime requires API, account, token, Harness, and state');
@@ -69,6 +91,7 @@ export class WeixinRuntime {
     this.#logger = logger;
     this.#replyTimeoutMs = replyTimeoutMs;
     this.#maxMessageChars = maxMessageChars;
+    this.#startRetryDelaysMs = startRetryDelays(startRetryDelaysMs);
   }
 
   get status() {
@@ -92,10 +115,7 @@ export class WeixinRuntime {
     try {
       await this.#harness.ensureRunning();
       this.#status.harnessReachable = true;
-      await this.#api.notifyStart({
-        baseUrl: this.#config.baseUrl,
-        token: this.#token,
-      });
+      await this.#notifyStart();
       this.#bridge = new WeixinHarnessBridge({
         api: this.#api,
         baseUrl: this.#config.baseUrl,
@@ -129,6 +149,25 @@ export class WeixinRuntime {
     }
   }
 
+  async #notifyStart() {
+    for (let attempt = 0; ; attempt += 1) {
+      try {
+        return await this.#api.notifyStart({
+          baseUrl: this.#config.baseUrl,
+          token: this.#token,
+        });
+      } catch (error) {
+        const wait = this.#startRetryDelaysMs[attempt];
+        if (wait === undefined || !retryableStartError(error)) throw error;
+        this.#logger.warn?.(
+          `[dsh-weixin] account ${this.#config.botId} start request failed; retrying in ${wait}ms:`,
+          error,
+        );
+        await delay(wait);
+      }
+    }
+  }
+
   async #runMonitor(signal) {
     let consecutiveFailures = 0;
     while (!signal.aborted) {

test/channels/weixin/weixin-runtime.test.mjs

@@ -1,6 +1,7 @@
 import assert from 'node:assert/strict';
 import test from 'node:test';
 
+import { WeixinApiError } from '../../../src/channels/weixin/weixin-api.mjs';
 import { WeixinRuntime } from '../../../src/channels/weixin/weixin-runtime.mjs';
 
 const flush = () => new Promise((resolve) => setImmediate(resolve));
@@ -89,3 +90,33 @@ test('runtime refuses to report ready when notifyStart rejects the stored token'
   assert.equal(runtime.status.ready, false);
   assert.equal(runtime.status.weixinConnectionState, 'failed');
 });
+
+test('runtime retries a transient notifyStart failure before reporting the account offline', async () => {
+  let startCalls = 0;
+  const runtime = new WeixinRuntime({
+    api: {
+      notifyStart: async () => {
+        startCalls += 1;
+        if (startCalls === 1) {
+          throw new WeixinApiError('network-error', 'temporary DNS failure');
+        }
+      },
+      notifyStop: async () => {},
+      sendText: async () => {},
+      getUpdates: async ({ signal }) => new Promise((_resolve, reject) => {
+        signal.addEventListener('abort', () => reject(new DOMException('Aborted', 'AbortError')));
+      }),
+    },
+    config: { botId: 'wx_retry', baseUrl: 'https://ilinkai.weixin.qq.com/', ownerUserId: 'owner' },
+    token: 'bot-token',
+    harness: { ensureRunning: async () => true },
+    state: {},
+    startRetryDelaysMs: [0],
+    logger: { warn() {}, error() {} },
+  });
+
+  const started = await runtime.start();
+  assert.equal(started.ready, true);
+  assert.equal(startCalls, 2);
+  await runtime.stop();
+});