From 63fba5d2fe1175d6a044ba5de21877b80f947cd7 Mon Sep 17 00:00:00 2001
From: nas <156536069+Nas01010101@users.noreply.github.com>
Date: Sat, 6 Jun 2026 12:06:31 -0400
Subject: [PATCH] fix(cron): require HTTP context for server_error retries

Narrow cron server_error retry classification so incidental 500-599 numbers in failure text no longer trigger retryable server_error. Genuine HTTP/status 5xx strings, canonical 5xx phrases, 5xx, and standalone terse codes still retry.

Maintainer proof: focused cron retry tests, formatter/lint/diff checks, clean autoreview, Testbox-through-Crabbox check:changed tbx_01kteteqqrppbzgh560sybe0nk / Actions run 27066938422, and green PR CI on 6124f14850eb57f393f36ea5e9bfe7ed5a60257b.

Fixes #90947.
---
 src/cron/retry-hint.test.ts | 34 ++++++++++++++++++++++++++++++++++
 src/cron/retry-hint.ts      | 13 ++++++++++++-
 2 files changed, 46 insertions(+), 1 deletion(-)

diff --git a/src/cron/retry-hint.test.ts b/src/cron/retry-hint.test.ts
index ff0324bbe84d..c2e4de25b1da 100644
--- a/src/cron/retry-hint.test.ts
+++ b/src/cron/retry-hint.test.ts
@@ -36,4 +36,38 @@ describe("resolveCronExecutionRetryHint", () => {
       retryable: false,
     });
   });
+
+  it("does not classify bare 5xx-looking numbers as server_error", () => {
+    for (const message of [
+      "context limit 512 exceeded",
+      "process exited with 503 lines of output",
+      "ENOENT: no such file '/var/run/app-540.sock'",
+      "killed worker pid 511 after deadline",
+      "assertion failed: expected 500 got 0",
+      "error 500 got 0",
+      "process exited with code 500",
+    ]) {
+      expect(resolveCronExecutionRetryHint(message, ["server_error"])).toEqual({
+        retryable: false,
+      });
+    }
+  });
+
+  it("classifies genuine HTTP 5xx errors as server_error", () => {
+    for (const message of [
+      "HTTP 503 Service Unavailable",
+      "received status 500 from upstream",
+      "500 Internal Server Error",
+      "502 Bad Gateway",
+      "upstream returned 5xx",
+      "response code: 502",
+      "503",
+      "500",
+    ]) {
+      expect(resolveCronExecutionRetryHint(message, ["server_error"])).toEqual({
+        retryable: true,
+        category: "server_error",
+      });
+    }
+  });
 });
diff --git a/src/cron/retry-hint.ts b/src/cron/retry-hint.ts
index 253265d5210a..1dc097e189ac 100644
--- a/src/cron/retry-hint.ts
+++ b/src/cron/retry-hint.ts
@@ -7,6 +7,17 @@ export type CronRetryHint = {
   category?: CronRetryOn;
 };
 
+// A bare 5xx-looking number embedded in prose is not an HTTP server error: cron
+// failure messages routinely contain such numbers ("context limit 512 exceeded",
+// "exited with 503 lines", "pid 511 killed", a "...-540.sock" path), and
+// /\b5\d{2}\b/ matched all of them, wrongly marking permanent failures retryable.
+// Match a 5xx number only with HTTP/status context, a canonical 5xx phrase, or
+// when it is the entire message (a terse "503"), so genuine
+// "500 Internal Server Error" / "502 Bad Gateway" / "5xx" still classify while
+// incidental numbers in longer messages do not.
+const SERVER_ERROR_PATTERN =
+  /\b(?:https?|status(?:[ _]code)?|response(?:[ _]code)?|http(?:[ _]status)?)\b[\s:=#"']{0,4}5\d{2}\b|\b5\d{2}\b[\s:)\].,-]*(?:internal server error|server error|bad gateway|service unavailable|gateway time-?out)\b|\binternal server error\b|\bbad gateway\b|\bservice unavailable\b|\bgateway time-?out\b|\b5xx\b|^\s*5\d{2}\s*$/i;
+
 const TRANSIENT_PATTERNS: Record<CronRetryOn, RegExp> = {
   rate_limit:
     /(rate[_ ]limit|too many requests|429|resource has been exhausted|cloudflare|tokens per day)/i,
@@ -15,7 +26,7 @@ const TRANSIENT_PATTERNS: Record<CronRetryOn, RegExp> = {
   network:
     /(network|fetch failed|socket|econnreset|econnrefused|eai_again|enetdown|ehostunreach|ehostdown|enetreset|enetunreach|epipe)/i,
   timeout: /(timeout|etimedout)/i,
-  server_error: /\b5\d{2}\b/,
+  server_error: SERVER_ERROR_PATTERN,
 };
 
 /** Classifies cron execution errors against the configured retryable transient categories. */