fix(gateway): classify wrapped "fetch failed" messages as transient network errors (openclaw#38530)

Verified:
- pnpm build
- pnpm check
- pnpm test:macmini

Co-authored-by: xinhuagu <562450+xinhuagu@users.noreply.github.com>
Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>

CHANGELOG.md

@@ -221,6 +221,7 @@ Docs: https://docs.openclaw.ai
 - Agents/skill API write pacing: add a global prompt guardrail that treats skill-driven external API writes as rate-limited by default, so runners prefer batched writes, avoid tight request loops, and respect `429`/`Retry-After`. Thanks @vincentkoc.
 - Google Chat/multi-account webhook auth fallback: when `channels.googlechat.accounts.default` carries shared webhook audience/path settings (for example after config normalization), inherit those defaults for named accounts while preserving top-level and per-account overrides, so inbound webhook verification no longer fails silently for named accounts missing duplicated audience fields. Fixes #38369.
 - Models/tool probing: raise the tool-capability probe budget from 32 to 256 tokens so reasoning models that spend tokens on thinking before returning a required tool call are less likely to be misclassified as not supporting tools. (#7521) Thanks @jakobdylanc.
+- Gateway/transient network classification: treat wrapped `...: fetch failed` transport messages as transient while avoiding broad matches like `Web fetch failed (404): ...`, preventing Discord reconnect wrappers from crashing the gateway without suppressing non-network tool failures. (#38530) Thanks @xinhuagu.
 
 ## 2026.3.2
 

src/infra/unhandled-rejections.fatal-detection.test.ts

@@ -86,7 +86,7 @@ describe("installUnhandledRejectionHandler - fatal detection", () => {
 
   describe("non-fatal errors", () => {
     it("does not exit on known transient network errors", () => {
-      const transientCases = [
+      const transientCases: unknown[] = [
         Object.assign(new TypeError("fetch failed"), {
           cause: { code: "UND_ERR_CONNECT_TIMEOUT", syscall: "connect" },
         }),
@@ -111,6 +111,11 @@ describe("installUnhandledRejectionHandler - fatal detection", () => {
         }),
       ];
 
+      // Wrapped fetch-failed (e.g. Discord: "Failed to get gateway information from Discord: fetch failed")
+      transientCases.push(
+        new Error("Failed to get gateway information from Discord: fetch failed"),
+      );
+
       for (const transientErr of transientCases) {
         expectExitCodeFromUnhandled(transientErr, []);
       }

src/infra/unhandled-rejections.test.ts

@@ -56,10 +56,13 @@ describe("isTransientNetworkError", () => {
       "EHOSTUNREACH",
       "ENETUNREACH",
       "EAI_AGAIN",
+      "EPROTO",
       "UND_ERR_CONNECT_TIMEOUT",
       "UND_ERR_SOCKET",
       "UND_ERR_HEADERS_TIMEOUT",
       "UND_ERR_BODY_TIMEOUT",
+      "ERR_SSL_WRONG_VERSION_NUMBER",
+      "ERR_SSL_PROTOCOL_RETURNED_AN_ERROR",
     ];
 
     for (const code of codes) {
@@ -122,6 +125,26 @@ describe("isTransientNetworkError", () => {
     expect(isTransientNetworkError(error)).toBe(true);
   });
 
+  it("returns true for wrapped fetch-failed messages from integration clients", () => {
+    const error = new Error("Failed to get gateway information from Discord: fetch failed");
+    expect(isTransientNetworkError(error)).toBe(true);
+  });
+
+  it("returns false for non-network fetch-failed wrappers from tools", () => {
+    const error = new Error("Web fetch failed (404): Not Found");
+    expect(isTransientNetworkError(error)).toBe(false);
+  });
+
+  it("returns true for TLS/SSL transient message snippets", () => {
+    expect(isTransientNetworkError(new Error("write EPROTO 00A8B0C9:error"))).toBe(true);
+    expect(
+      isTransientNetworkError(
+        new Error("SSL routines:OPENSSL_internal:WRONG_VERSION_NUMBER while connecting"),
+      ),
+    ).toBe(true);
+    expect(isTransientNetworkError(new Error("tlsv1 alert protocol version"))).toBe(true);
+  });
+
   it("returns false for regular errors without network codes", () => {
     expect(isTransientNetworkError(new Error("Something went wrong"))).toBe(false);
     expect(isTransientNetworkError(new TypeError("Cannot read property"))).toBe(false);

src/infra/unhandled-rejections.ts

@@ -38,6 +38,9 @@ const TRANSIENT_NETWORK_CODES = new Set([
   "UND_ERR_SOCKET",
   "UND_ERR_HEADERS_TIMEOUT",
   "UND_ERR_BODY_TIMEOUT",
+  "EPROTO",
+  "ERR_SSL_WRONG_VERSION_NUMBER",
+  "ERR_SSL_PROTOCOL_RETURNED_AN_ERROR",
 ]);
 
 const TRANSIENT_NETWORK_ERROR_NAMES = new Set([
@@ -49,7 +52,7 @@ const TRANSIENT_NETWORK_ERROR_NAMES = new Set([
 ]);
 
 const TRANSIENT_NETWORK_MESSAGE_CODE_RE =
-  /\b(ECONNRESET|ECONNREFUSED|ENOTFOUND|ETIMEDOUT|ESOCKETTIMEDOUT|ECONNABORTED|EPIPE|EHOSTUNREACH|ENETUNREACH|EAI_AGAIN|UND_ERR_CONNECT_TIMEOUT|UND_ERR_DNS_RESOLVE_FAILED|UND_ERR_CONNECT|UND_ERR_SOCKET|UND_ERR_HEADERS_TIMEOUT|UND_ERR_BODY_TIMEOUT)\b/i;
+  /\b(ECONNRESET|ECONNREFUSED|ENOTFOUND|ETIMEDOUT|ESOCKETTIMEDOUT|ECONNABORTED|EPIPE|EHOSTUNREACH|ENETUNREACH|EAI_AGAIN|EPROTO|UND_ERR_CONNECT_TIMEOUT|UND_ERR_DNS_RESOLVE_FAILED|UND_ERR_CONNECT|UND_ERR_SOCKET|UND_ERR_HEADERS_TIMEOUT|UND_ERR_BODY_TIMEOUT)\b/i;
 
 const TRANSIENT_NETWORK_MESSAGE_SNIPPETS = [
   "getaddrinfo",
@@ -58,8 +61,22 @@ const TRANSIENT_NETWORK_MESSAGE_SNIPPETS = [
   "network error",
   "network is unreachable",
   "temporary failure in name resolution",
+  "tlsv1 alert",
+  "ssl routines",
+  "packet length too long",
+  "write eproto",
 ];
 
+function isWrappedFetchFailedMessage(message: string): boolean {
+  if (message === "fetch failed") {
+    return true;
+  }
+
+  // Keep wrapped variants (for example "...: fetch failed") while avoiding broad
+  // matches like "Web fetch failed (404): ..." that are not transport failures.
+  return /:\s*fetch failed$/.test(message);
+}
+
 function getErrorCause(err: unknown): unknown {
   if (!err || typeof err !== "object") {
     return undefined;
@@ -154,10 +171,6 @@ export function isTransientNetworkError(err: unknown): boolean {
       return true;
     }
 
-    if (candidate instanceof TypeError && candidate.message === "fetch failed") {
-      return true;
-    }
-
     if (!candidate || typeof candidate !== "object") {
       continue;
     }
@@ -169,7 +182,7 @@ export function isTransientNetworkError(err: unknown): boolean {
     if (TRANSIENT_NETWORK_MESSAGE_CODE_RE.test(message)) {
       return true;
     }
-    if (message === "fetch failed") {
+    if (isWrappedFetchFailedMessage(message)) {
       return true;
     }
     if (TRANSIENT_NETWORK_MESSAGE_SNIPPETS.some((snippet) => message.includes(snippet))) {