From de04e21aa14096edb1e417b35c26a248594eb814 Mon Sep 17 00:00:00 2001
From: zhi <zhi@hangman-lab.top>
Date: Thu, 26 Feb 2026 08:45:37 +0000
Subject: [PATCH] fix: wait for gateway ready before post-install model
 validation

Root cause: gateway restart is async (systemd), but validateNoReplyModelAvailable()
ran immediately after, hitting a race condition where the new gateway process
hadn't finished initializing yet. This caused 'model not listed' validation
failures, triggering config rollback even though the config was correct.

Changes:
- Add waitForGatewayReady() that polls 'openclaw gateway status' for RPC probe
- Add retry loop (5 attempts, 2s interval) to validateNoReplyModelAvailable()
- Fix CONFIG.example.json: contextWindow 4096->200000, maxTokens 64->8192
  (OpenClaw requires minimum 16000 contextWindow)
---
 docs/CONFIG.example.json                 |  4 +-
 no-reply-api/package-lock.json           | 12 ++++++
 scripts/install-whispergate-openclaw.mjs | 48 +++++++++++++++++++-----
 3 files changed, 53 insertions(+), 11 deletions(-)
 create mode 100644 no-reply-api/package-lock.json

diff --git a/docs/CONFIG.example.json b/docs/CONFIG.example.json
index 0c887fb..d14f120 100644
--- a/docs/CONFIG.example.json
+++ b/docs/CONFIG.example.json
@@ -40,8 +40,8 @@
             "reasoning": false,
             "input": ["text"],
             "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
-            "contextWindow": 4096,
-            "maxTokens": 64
+            "contextWindow": 200000,
+            "maxTokens": 8192
           }
         ]
       }
diff --git a/no-reply-api/package-lock.json b/no-reply-api/package-lock.json
new file mode 100644
index 0000000..3e29cb5
--- /dev/null
+++ b/no-reply-api/package-lock.json
@@ -0,0 +1,12 @@
+{
+  "name": "whispergate-no-reply-api",
+  "version": "0.1.0",
+  "lockfileVersion": 3,
+  "requires": true,
+  "packages": {
+    "": {
+      "name": "whispergate-no-reply-api",
+      "version": "0.1.0"
+    }
+  }
+}
diff --git a/scripts/install-whispergate-openclaw.mjs b/scripts/install-whispergate-openclaw.mjs
index 7cc3e7b..5adf132 100755
--- a/scripts/install-whispergate-openclaw.mjs
+++ b/scripts/install-whispergate-openclaw.mjs
@@ -45,17 +45,41 @@ function runOpenclaw(args, { allowFail = false } = {}) {
   }
 }
 
-function validateNoReplyModelAvailable() {
+function sleep(ms) {
+  return new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+async function waitForGatewayReady(maxWaitMs = 30000) {
+  const start = Date.now();
+  const interval = 1500;
+  while (Date.now() - start < maxWaitMs) {
+    const probe = runOpenclaw(["gateway", "status"], { allowFail: true }) || "";
+    if (probe.includes("RPC probe: ok")) return true;
+    console.log(`[whispergate] waiting for gateway ready... (${Math.round((Date.now() - start) / 1000)}s)`);
+    await sleep(interval);
+  }
+  return false;
+}
+
+async function validateNoReplyModelAvailable(retries = 5, delayMs = 2000) {
   const modelRef = `${NO_REPLY_PROVIDER_ID}/${NO_REPLY_MODEL_ID}`;
-  const list = runOpenclaw(["models", "list"], { allowFail: true }) || "";
-  if (!list.includes(modelRef)) {
-    throw new Error(`post-install validation failed: model not listed: ${modelRef}`);
+
+  for (let attempt = 1; attempt <= retries; attempt++) {
+    const list = runOpenclaw(["models", "list"], { allowFail: true }) || "";
+    if (list.includes(modelRef)) {
+      const status = runOpenclaw(["models", "status", "--json"], { allowFail: true }) || "";
+      if (status.includes(NO_REPLY_PROVIDER_ID)) {
+        console.log(`[whispergate] model validation passed on attempt ${attempt}`);
+        return;
+      }
+    }
+    if (attempt < retries) {
+      console.log(`[whispergate] model not yet visible (attempt ${attempt}/${retries}), retrying in ${delayMs}ms...`);
+      await sleep(delayMs);
+    }
   }
 
-  const status = runOpenclaw(["models", "status", "--json"], { allowFail: true }) || "";
-  if (!status.includes(NO_REPLY_PROVIDER_ID)) {
-    throw new Error(`post-install validation failed: provider not visible in models status: ${NO_REPLY_PROVIDER_ID}`);
-  }
+  throw new Error(`post-install validation failed: model not listed after ${retries} attempts: ${modelRef}`);
 }
 
 function getJson(pathKey) {
@@ -175,7 +199,13 @@ if (mode === "install") {
     setJson(PATH_PROVIDERS, providers);
 
     runOpenclaw(["gateway", "restart"]);
-    validateNoReplyModelAvailable();
+    console.log("[whispergate] gateway restart issued, waiting for ready...");
+    const ready = await waitForGatewayReady();
+    if (!ready) {
+      throw new Error("gateway did not become ready within 30s after restart");
+    }
+    console.log("[whispergate] gateway ready, validating model...");
+    await validateNoReplyModelAvailable();
 
     const after = {
       [PATH_PLUGINS_LOAD]: getJson(PATH_PLUGINS_LOAD),