retry capacity-style upstream errors

2026-06-29 15:54:36 +08:00
parent 778068af42
commit 283c14b6b4
8 changed files with 384 additions and 24 deletions
@@ -123,6 +123,20 @@ function startFakeUpstream(port) {
          ]);
          return;
        }
+        if (parsed.test_capacity_error) {
+          createJsonResponse(
+            res,
+            parsed.test_capacity_status ?? 503,
+            {
+              error: {
+                message: parsed.test_capacity_message || "Selected model is at capacity. Please try a different model.",
+                type: "server_error",
+              },
+            },
+            { "x-upstream-test": "capacity-error" },
+          );
+          return;
+        }
        if (parsed.stream) {
          createSseResponse(res, [
            'data: {"type":"response.output_text.delta","delta":"hello"}\n\n',
@@ -273,11 +287,13 @@ async function run() {
    listen_port: gatewayPort,
    upstream_base_url: `http://127.0.0.1:${upstreamPort}`,
    request_body_limit_bytes: 10 * 1024 * 1024,
-    endpoints: ["/responses", "/chat/completions", "/v1/responses", "/v1/chat/completions"],
-    reasoning_equals: [516],
-    non_stream_status_code: 502,
+    endpoints: ["/responses", "/chat/completions", "/v1/responses", "/v1/chat/completions"],
+    reasoning_equals: [516],
+    retryable_status_codes: [429, 503],
+    retryable_error_messages: ["Selected model is at capacity. Please try a different model."],
+    non_stream_status_code: 502,
    stream_action: "strict_502",
-    log_match: true,
+    log_match: true,
    health_path: "/__codex_retry_gateway/health",
  };

@@ -349,6 +365,34 @@ async function run() {
      `请求体大小记录异常: ${recoveredEntry?.request_body_bytes}`,
    );

+    const capacityResponse = await fetch(`http://127.0.0.1:${gatewayPort}/responses`, {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ test_capacity_error: true }),
+    });
+    const capacityBody = await capacityResponse.json();
+    assert(capacityResponse.status === 502, `capacity error 未返回 502: ${capacityResponse.status}`);
+    assert(
+      capacityBody?.error?.code === "upstream_error_retry_triggered",
+      "capacity error 返回体未标记 retry trigger",
+    );
+    assert(
+      capacityBody?.error?.upstream_status_code === 503,
+      "capacity error 返回体未保留 upstream status",
+    );
+
+    const streamCapacityResponse = await fetch(`http://127.0.0.1:${gatewayPort}/responses`, {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ stream: true, test_capacity_error: true }),
+    });
+    const streamCapacityBody = await streamCapacityResponse.json();
+    assert(streamCapacityResponse.status === 502, `stream+capacity error 未返回 502: ${streamCapacityResponse.status}`);
+    assert(
+      streamCapacityBody?.error?.code === "upstream_error_retry_triggered",
+      "stream+capacity error 返回体未标记 retry trigger",
+    );
+
    for (const streamPath of [
      "/responses",
      "/v1/responses",