7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
TypeScript
fix(router): treat 413 Payload Too Large as retryable (#64, credits @danscMax)
danscMax839fe5a56092
diff --git a/server/src/__tests__/routes/proxy-retry.test.ts b/server/src/__tests__/routes/proxy-retry.test.tsnew file mode 100644index 00000000..76784712--- /dev/null+++ b/server/src/__tests__/routes/proxy-retry.test.ts@@ -0,0 +1,40 @@+import { describe, it, expect } from 'vitest';+import { isRetryableError } from '../../routes/proxy.js';++describe('isRetryableError', () => {+ describe('413 Payload Too Large (the bug this PR fixes)', () => {+ it('treats explicit "413" in the error message as retryable', () => {+ expect(isRetryableError(new Error('GitHub Models API error 413: Request body too large'))).toBe(true);+ expect(isRetryableError(new Error('Cloudflare API error 413: Payload Too Large'))).toBe(true);+ });++ it('treats common 413 phrasings (no status code) as retryable', () => {+ expect(isRetryableError(new Error('Payload Too Large'))).toBe(true);+ expect(isRetryableError(new Error('Request body too large for this model'))).toBe(true);+ expect(isRetryableError(new Error('Request entity too large'))).toBe(true);+ expect(isRetryableError(new Error('Content too large'))).toBe(true);+ });+ });++ describe('existing categories still classify correctly', () => {+ it('429 / rate limits are retryable', () => {+ expect(isRetryableError(new Error('429 Too Many Requests'))).toBe(true);+ expect(isRetryableError(new Error('rate limit exceeded'))).toBe(true);+ expect(isRetryableError(new Error('quota exhausted'))).toBe(true);+ });++ it('5xx and network errors are retryable', () => {+ expect(isRetryableError(new Error('503 Service Unavailable'))).toBe(true);+ expect(isRetryableError(new Error('500 Internal Server Error'))).toBe(true);+ expect(isRetryableError(new Error('ETIMEDOUT'))).toBe(true);+ expect(isRetryableError(new Error('ECONNREFUSED'))).toBe(true);+ });++ it('4xx other than 413/429 are NOT retryable', () => {+ expect(isRetryableError(new Error('401 Unauthorized'))).toBe(false);+ expect(isRetryableError(new Error('403 Forbidden'))).toBe(false);+ expect(isRetryableError(new Error('404 Not Found'))).toBe(false);+ expect(isRetryableError(new Error('400 Bad Request'))).toBe(false);+ });+ });+});diff --git a/server/src/routes/proxy.ts b/server/src/routes/proxy.tsindex b6eb772f..d60a86ed 100644--- a/server/src/routes/proxy.ts+++ b/server/src/routes/proxy.ts@@ -189,14 +189,18 @@ const chatCompletionSchema = z.object({parallel_tool_calls: z.boolean().optional(),});-function isRetryableError(err: any): boolean {+export function isRetryableError(err: any): boolean {const msg = (err.message ?? '').toLowerCase();return msg.includes('429') || msg.includes('rate limit') || msg.includes('too many requests')|| msg.includes('quota') || msg.includes('resource_exhausted')|| msg.includes('aborted') || msg.includes('timeout') || msg.includes('etimedout')|| msg.includes('econnrefused') || msg.includes('econnreset')|| msg.includes('503') || msg.includes('unavailable')- || msg.includes('500') || msg.includes('internal server error');+ || msg.includes('500') || msg.includes('internal server error')+ // 413: this model's payload limit is too small for the request, but another+ // provider in the fallback chain may have a larger limit. Same reasoning as 503.+ || msg.includes('413') || msg.includes('payload too large') || msg.includes('request body too large')+ || msg.includes('request entity too large') || msg.includes('content too large');}proxyRouter.post('/chat/completions', async (req: Request, res: Response) => {