fix(failover): classify INTERNAL 500 responses as retryable timeouts (#68238)

* Agents: treat Google INTERNAL 500 as timeout failover

(cherry picked from commit c2538523a22d39b65c6b4056ab4857ee84f06887)

* test(failover): narrow INTERNAL timeout patterns

* fix: document INTERNAL timeout retry guard

* fix: ignore plain status prose in server error classification

* fix(failover): preserve mixed server-error retry signals

* test(failover): dedupe internal status samples

* fix(failover): retry status prose with code 500

* fix: classify INTERNAL 500 responses as retryable timeouts

* fix: classify INTERNAL 500 responses as retryable timeouts

---------

Co-authored-by: Kosbling <github@kosbling.com>
Co-authored-by: Openbling <github@openbling.ai>

CHANGELOG.md

@@ -34,6 +34,7 @@ Docs: https://docs.openclaw.ai
 - Feishu/card actions: resolve card-action chat type from the Feishu chat API when stored context is missing, preferring `chat_mode` over `chat_type`, so DM-originated card actions no longer bypass `dmPolicy` by falling through to the group handling path. (#68201)
 - Cron/isolated-agent: preserve `trusted: false` on isolated cron awareness events mirrored into the main session, and forward the optional `trusted` flag through the gateway cron wrapper so explicit trust downgrades survive session-key scoping. (#68210)
 - Agents/fallback: recognize bare leading ZenMux `402 ...` quota-refresh errors without misclassifying plain numeric `402 ...` text, and keep the embedded fallback regression coverage stable. (#47579) Thanks @bwjoke.
+- Failover/google: only treat `INTERNAL` status payloads as retryable timeouts when they also carry a `500` code, so malformed non-500 payloads do not enter the retry path. (#68238) Thanks @altaywtf and @Openbling.
 
 ## 2026.4.15
 

src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts

@@ -45,6 +45,12 @@ const GROQ_TOO_MANY_REQUESTS_MESSAGE =
   "429 Too Many Requests: Too many requests were sent in a given timeframe.";
 const GROQ_SERVICE_UNAVAILABLE_MESSAGE =
   "503 Service Unavailable: The server is temporarily unable to handle the request due to overloading or maintenance."; // pragma: allowlist secret
+const PLAIN_INTERNAL_SERVER_ERROR_STATUS_SAMPLE =
+  "Proxy notice: Status: Internal Server Error";
+const MIXED_INTERNAL_SERVER_ERROR_STATUS_SAMPLE =
+  `${PLAIN_INTERNAL_SERVER_ERROR_STATUS_SAMPLE}; upstream connect error`;
+const INTERNAL_SERVER_ERROR_STATUS_WITH_500_SAMPLE =
+  `${PLAIN_INTERNAL_SERVER_ERROR_STATUS_SAMPLE}; code:500`;
 
 function expectMessageMatches(
   matcher: (message: string) => boolean,
@@ -64,6 +70,12 @@ function expectTimeoutFailoverSamples(samples: readonly string[]) {
   }
 }
 
+function expectNotFailoverSample(sample: string) {
+  expect(isTimeoutErrorMessage(sample)).toBe(false);
+  expect(classifyFailoverReason(sample)).toBeNull();
+  expect(isFailoverErrorMessage(sample)).toBe(false);
+}
+
 describe("isAuthPermanentErrorMessage", () => {
   it.each([
     {
@@ -811,6 +823,30 @@ describe("isFailoverErrorMessage", () => {
     expect(classifyFailoverReason(sample)).toBe(null);
     expect(isFailoverErrorMessage(sample)).toBe(false);
   });
+
+  it("matches google INTERNAL status errors as timeout", () => {
+    const sample =
+      "provider=google model=gemini-3.1-flash-lite-preview got status: INTERNAL upstream failure code:500";
+    expect(isTimeoutErrorMessage(sample)).toBe(true);
+    expect(classifyFailoverReason(sample)).toBe("timeout");
+    expect(isFailoverErrorMessage(sample)).toBe(true);
+  });
+
+  it("does not treat plain status text with internal-server-error wording as timeout", () => {
+    expectNotFailoverSample(PLAIN_INTERNAL_SERVER_ERROR_STATUS_SAMPLE);
+  });
+
+  it("keeps mixed upstream server errors retryable when they also mention status prose", () => {
+    expect(isTimeoutErrorMessage(MIXED_INTERNAL_SERVER_ERROR_STATUS_SAMPLE)).toBe(false);
+    expect(classifyFailoverReason(MIXED_INTERNAL_SERVER_ERROR_STATUS_SAMPLE)).toBe("timeout");
+    expect(isFailoverErrorMessage(MIXED_INTERNAL_SERVER_ERROR_STATUS_SAMPLE)).toBe(true);
+  });
+
+  it("keeps status prose retryable when it is explicitly paired with code 500", () => {
+    expect(isTimeoutErrorMessage(INTERNAL_SERVER_ERROR_STATUS_WITH_500_SAMPLE)).toBe(false);
+    expect(classifyFailoverReason(INTERNAL_SERVER_ERROR_STATUS_WITH_500_SAMPLE)).toBe("timeout");
+    expect(isFailoverErrorMessage(INTERNAL_SERVER_ERROR_STATUS_WITH_500_SAMPLE)).toBe(true);
+  });
 });
 
 describe("parseImageSizeError", () => {
@@ -1230,4 +1266,33 @@ describe("classifyProviderRuntimeFailureKind", () => {
       ),
     ).not.toBe("proxy");
   });
+
+  it("classifies google-style INTERNAL status payloads as timeout", () => {
+    expect(
+      classifyFailoverReason(
+        'ERROR provider=google model=gemini-3.1-flash-lite-preview: got status: INTERNAL, details: {"code":500,"status":"INTERNAL"}',
+      ),
+    ).toBe("timeout");
+    expect(
+      classifyFailoverReason(
+        'got status: INTERNAL. {"error":{"code":500,"message":"Internal error encountered.","status":"INTERNAL"}}',
+      ),
+    ).toBe("timeout");
+  });
+
+  it("does not classify google-style INTERNAL payloads without a 500 code as timeout", () => {
+    const sample =
+      'got status: INTERNAL. {"error":{"code":400,"message":"Request malformed","status":"INTERNAL"}}';
+    expect(isTimeoutErrorMessage(sample)).toBe(false);
+    expect(classifyFailoverReason(sample)).toBeNull();
+    expect(isFailoverErrorMessage(sample)).toBe(false);
+  });
+
+  it("does not classify plain status text with internal server error wording as timeout", () => {
+    expectNotFailoverSample(PLAIN_INTERNAL_SERVER_ERROR_STATUS_SAMPLE);
+  });
+
+  it("classifies internal server error status prose with code 500 as timeout", () => {
+    expect(classifyFailoverReason(INTERNAL_SERVER_ERROR_STATUS_WITH_500_SAMPLE)).toBe("timeout");
+  });
 });

src/agents/pi-embedded-helpers/failover-matches.ts

@@ -42,6 +42,9 @@ const COMMON_AUTH_ERROR_PATTERNS = [
 
 const ZAI_BILLING_CODE_1311_RE = /"code"\s*:\s*1311\b/;
 const ZAI_AUTH_CODE_1113_RE = /"code"\s*:\s*1113\b/;
+const STATUS_INTERNAL_SERVER_ERROR_RE = /\bstatus:\s*internal server error\b/i;
+const STATUS_INTERNAL_SERVER_ERROR_WITH_500_RE =
+  /^(?=[\s\S]*\bstatus:\s*internal server error\b)(?=[\s\S]*\bcode["']?\s*[:=]\s*500\b)/i;
 
 const ZAI_AUTH_ERROR_PATTERNS = [
   // Z.ai: error 1113 = wrong endpoint or invalid credentials (#48988)
@@ -95,6 +98,8 @@ const ERROR_PATTERNS = {
     "service unavailable",
     "deadline exceeded",
     "context deadline exceeded",
+    /^(?=[\s\S]*\bgot status:\s*internal\b)(?=[\s\S]*\bcode["']?\s*[:=]\s*500\b)/i,
+    /^(?=[\s\S]*["']status["']\s*:\s*["']internal["'])(?=[\s\S]*["']code["']\s*:\s*500\b)/i,
     "connection error",
     "network error",
     "network request failed",
@@ -233,5 +238,13 @@ export function isOverloadedErrorMessage(raw: string): boolean {
 }
 
 export function isServerErrorMessage(raw: string): boolean {
-  return matchesErrorPatterns(raw, ERROR_PATTERNS.serverError);
+  const value = normalizeLowercaseStringOrEmpty(raw);
+  if (!value) {
+    return false;
+  }
+  if (STATUS_INTERNAL_SERVER_ERROR_WITH_500_RE.test(value)) {
+    return true;
+  }
+  const scrubbed = value.replace(STATUS_INTERNAL_SERVER_ERROR_RE, "").trim();
+  return scrubbed.length > 0 && matchesErrorPatterns(scrubbed, ERROR_PATTERNS.serverError);
 }