ghc-proxy 0.8.0 → 0.8.1

package/README.md

@@ -121,9 +121,9 @@ bunx ghc-proxy@latest selfcheck      # Probe the packaged bundle (loads every to
 | `--idle-timeout` | -- | `120` | Bun server idle timeout in seconds (`0` disables; Bun max is `255`; streaming routes disable idle timeout automatically) |
 | `--upstream-timeout` | -- | `1800` | Upstream request timeout in seconds (0 to disable) |
 | `--upstream-queue-concurrency` | -- | `10` | Maximum concurrent Copilot upstream requests |
-| `--upstream-queue-retries` | -- | `5` | Maximum retries for upstream 429 responses |
-| `--upstream-queue-base-delay` | -- | `2` | Base delay in seconds for upstream 429 backoff when `Retry-After` is absent |
-| `--upstream-queue-max-delay` | -- | `60` | Maximum delay in seconds for upstream 429 backoff |
+| `--upstream-queue-retries` | -- | `5` | Maximum retries for transient upstream responses (`408`, `429`, `500`, `502`, `503`, `504`, `529`) |
+| `--upstream-queue-base-delay` | -- | `2` | Base delay in seconds for upstream retry backoff when `Retry-After` is absent |
+| `--upstream-queue-max-delay` | -- | `60` | Maximum delay in seconds for upstream retry backoff |
 | `--ghe-domain` | `--ghe` | -- | GitHub Enterprise Cloud company domain (e.g. `company.ghe.com`). Required for GHE.com device login on first run; persisted automatically for later runs. |
 
 ## Rate Limiting
@@ -206,9 +206,9 @@ All fields are optional. The full schema:
 | `responsesOfficialEmulatorTtlSeconds` | `number` | `14400` | In-memory TTL for locally emulated Responses state |
 | `modelReasoningEfforts` | `Record<string, string>` | -- | Per-model reasoning effort defaults for Anthropic-to-Responses translation. Each value must be one of `none`, `minimal`, `low`, `medium`, `high`, or `xhigh` |
 | `upstreamQueueConcurrency` | `number` | `10` | Maximum concurrent Copilot upstream requests |
-| `upstreamQueueMaxRetries` | `number` | `5` | Maximum retries for upstream 429 responses |
-| `upstreamQueueBaseDelaySeconds` | `number` | `2` | Base delay (seconds) for upstream 429 backoff when `Retry-After` is absent |
-| `upstreamQueueMaxDelaySeconds` | `number` | `60` | Maximum delay (seconds) for upstream 429 backoff |
+| `upstreamQueueMaxRetries` | `number` | `5` | Maximum retries for transient upstream responses (`408`, `429`, `500`, `502`, `503`, `504`, `529`) |
+| `upstreamQueueBaseDelaySeconds` | `number` | `2` | Base delay (seconds) for upstream retry backoff when `Retry-After` is absent |
+| `upstreamQueueMaxDelaySeconds` | `number` | `60` | Maximum delay (seconds) for upstream retry backoff |
 | `gheDomain` | `string` | -- | GitHub Enterprise Cloud company domain (persisted automatically after GHE.com auth) |
 
 Example:

package/dist/main.mjs

@@ -5980,6 +5980,31 @@ var HTTPError = class extends Error {
 		return Response.json(this.body, { status: this.status });
 	}
 };
+const TRANSIENT_UPSTREAM_STATUSES = new Set([
+	408,
+	429,
+	500,
+	502,
+	503,
+	504,
+	529
+]);
+/**
+* Upstream statuses worth retrying: request-scoped faults (timeouts, gateway
+* errors) plus capacity limits. Shared by `UpstreamRequestQueue` and the
+* Copilot token refresh so both agree on what "transient" means.
+*/
+function isTransientUpstreamStatus(status) {
+	return TRANSIENT_UPSTREAM_STATUSES.has(status);
+}
+/**
+* Capacity signals that apply to the whole account or service rather than one
+* request. Only these warrant global back-pressure — applying a queue-wide
+* cooldown to a request-scoped 5xx turns one bad request into a proxy stall.
+*/
+function isCapacityLimitStatus(status) {
+	return status === 429 || status === 529;
+}
 function throwInvalidRequestError(message, param, code) {
 	throw new HTTPError(400, { error: {
 		message,
@@ -6014,7 +6039,9 @@ function isStructuredErrorPayload(value) {
 	return typeof value === "object" && value !== null && "error" in value && typeof value.error === "object" && value.error !== null;
 }
 function upstreamErrorType(status) {
-	return status === 429 ? "rate_limit_error" : "upstream_error";
+	if (status === 429) return "rate_limit_error";
+	if (status === 529) return "overloaded_error";
+	return "upstream_error";
 }
 function createFallbackUpstreamError(message, response, rawText) {
 	return { error: {
@@ -6957,8 +6984,10 @@ var UpstreamRequestQueue = class {
 				lease.release();
 				throw error;
 			}
-			if (response.status !== 429 || !context.retryable || attempt >= this.options.maxRetries) {
-				if (response.status === 429 && !context.retryable) this.applyCooldown(this.getRetryDelayMs(response, 0));
+			const { status } = response;
+			const isCapacityLimit = isCapacityLimitStatus(status);
+			if (!(isTransientUpstreamStatus(status) && context.retryable === true && attempt < this.options.maxRetries)) {
+				if (isCapacityLimit) this.applyCooldown(this.getRetryDelayMs(response, 0));
 				return {
 					response,
 					release: lease.release
@@ -6966,10 +6995,10 @@ var UpstreamRequestQueue = class {
 			}
 			const delayMs = this.getRetryDelayMs(response, attempt);
 			await discardResponse(response);
+			if (isCapacityLimit) this.applyCooldown(delayMs);
 			lease.release();
-			this.applyCooldown(delayMs);
 			this.logger.warn([
-				"Upstream rate limited;",
+				`Upstream ${status};`,
 				`retrying ${formatRequestContext(context)}`,
 				`in ${formatDurationMs(delayMs)}`,
 				`(attempt ${attempt + 1}/${this.options.maxRetries})`
@@ -7244,16 +7273,8 @@ async function setupGitHubToken(options) {
 function isAuthError(error) {
 	return error instanceof HTTPError && (error.status === 401 || error.status === 403);
 }
-const TRANSIENT_HTTP_STATUSES = new Set([
-	408,
-	429,
-	500,
-	502,
-	503,
-	504
-]);
 function isTransientHttpError(error) {
-	return TRANSIENT_HTTP_STATUSES.has(error.status);
+	return isTransientUpstreamStatus(error.status);
 }
 async function logUser() {
 	const user = await createGitHubClient().getGitHubUser();
@@ -7364,7 +7385,7 @@ const checkUsage = defineCommand({
 });
 //#endregion
 //#region src/util/version.ts
-const VERSION = "0.8.0";
+const VERSION = "0.8.1";
 //#endregion
 //#region src/debug.ts
 function getRuntimeInfo() {
@@ -53194,15 +53215,15 @@ const start = defineCommand({
 		},
 		"upstream-queue-retries": {
 			type: "string",
-			description: "Maximum retries for upstream 429 responses (default: 5)"
+			description: "Maximum retries for transient upstream responses (default: 5)"
 		},
 		"upstream-queue-base-delay": {
 			type: "string",
-			description: "Base delay in seconds for upstream 429 backoff when Retry-After is absent (default: 2)"
+			description: "Base delay in seconds for upstream retry backoff when Retry-After is absent (default: 2)"
 		},
 		"upstream-queue-max-delay": {
 			type: "string",
-			description: "Maximum delay in seconds for upstream 429 backoff (default: 60)"
+			description: "Maximum delay in seconds for upstream retry backoff (default: 60)"
 		},
 		"ghe-domain": {
 			alias: "ghe",

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "ghc-proxy",
   "type": "module",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "GitHub Copilot to OpenAI/Anthropic API proxy - Use Copilot with Claude Code, Cursor, and more",
   "author": "wxxb789 <wxxb789@outlook.com>",
   "homepage": "https://github.com/wxxb789/ghc-proxy",
@@ -41,6 +41,7 @@
     "smoke:cache": "bun run scripts/smoke/cache-control.ts",
     "smoke:packaged": "bun run scripts/smoke/packaged-cli.ts",
     "smoke:translation": "bun run scripts/smoke/translation.ts",
+    "serve": "NODE_ENV=production bun run ./src/main.ts start --wait --idle-timeout=255 --upstream-timeout=1800",
     "start": "NODE_ENV=production bun run ./src/main.ts",
     "typecheck": "npx tsc --project tsconfig.json"
   },