Latest archive
Recently updated troubleshooting pages, kept as a crawlable archive so new and refreshed fixes stay easy to discover.
Fix quota exceeded error on paid Gemini API tier — developers have active billing and minimal usage but are stuck on free-tier quotas Includes evidence for Google Gemini API troubleshooting demand.
"Quota exceeded. Please try again later." on Google AI Studio despite being a paid Tier 1 project with minimal usage and active billing. Web developers whose sites are protected by Cloudflare experience intermittent 522 errors preventing visitors from reaching their origin servers; need diagnostic steps and resolution methods Includes evidence for Cloudflare troubleshooting demand.
Error 522 Origin connection timed out — Cloudflare Workers unable to reach upstream origin Fix Claude Code OAuth login failures caused by DNS resolution failure of auth.anthropic.com Includes evidence for Claude Code troubleshooting demand.
getaddrinfo ENOTFOUND auth.anthropic.com — Claude Code OAuth login fails with timeout; auth.anthropic.com does not resolve via DNS, making authentication impossible Resolve Docker Compose or docker run port binding conflicts where host port is already occupied by another container or host process Includes evidence for Docker troubleshooting demand.
ERROR: Bind for 0.0.0.0:5001 failed: port is already allocated OR bind: address already in use Fix GitHub Actions workflow permission errors including GITHUB_TOKEN scope issues, fork PR restrictions, and runner filesystem ownership conflicts from root vs non-root step runs Includes evidence for GitHub Actions troubleshooting demand.
Error: HttpError: Resource not accessible by integration OR remote: Permission to user/repo.git denied to github-actions[bot] OR Error: EACCES: permission denied, open '/home/runner/.npm/_cacache/...' Diagnose and fix GitHub Copilot CLI authentication failures caused by silent swallowing of REST quota exceeded and rate limit errors during token validation phase Includes evidence for GitHub Copilot troubleshooting demand.
Session was not created with authentication info or custom provider (root cause: CLI auth flow silently swallows REST quota/rate-limit errors during token validation) Fix LiteLLM proxy ConnectionRefusedError by verifying proxy process is running on expected host:port, resolving Docker network isolation, unsetting conflicting LITELLM_PORT env variable Includes evidence for LiteLLM troubleshooting demand.
litellm.llms.custom_httpx.http_handler.HTTPHandler.post: Connection refused OR httpx.ConnectError: [Errno 111] Connection refused Fix OpenAI 429 rate limit error by understanding RPM/TPM dimensions and configuring proper retry/backoff strategy Includes evidence for OpenAI API troubleshooting demand.
Rate limit reached for gpt-3.5-turbo in organization org-exampleorgid123 on tokens per min. Limit: 10000.000000 / min. Current: 10020.000000 / min. Fix Vercel deployment build failures caused by environment drift (Node version mismatch, missing build-time env vars like NEXT_PUBLIC_*), TypeScript stricter compilation, or OOM conditions Includes evidence for Vercel troubleshooting demand.
Cannot find module OR "Module not found" OR Process completed with exit code 1 OR Vercel build failed despite local npm run build being green User configures an MCP server in Claude Code and receives ECONNREFUSED because the server hasn't finished listening yet; needs fix for startup race condition and localhost IPv6 binding. Includes evidence for Claude Code troubleshooting demand.
Error: connect ECONNREFUSED 127.0.0.1:3000 / TCPConnectWrap.afterConnect [as oncomplete] After upgrading to LiteLLM 1.82.1, previously working OpenRouter model IDs throw 404 errors. Users proxying through LiteLLM proxy lose access to OpenRouter models and need quick workaround to restore routing to healthy deployments. Includes evidence for LiteLLM troubleshooting demand.
litellm.CallFailed: Error code: 404 — OpenRouter model IDs stopped working after LiteLLM upgrade to v1.82.1 Developers using GitHub Copilot CLI via MCP sessions experience silent OAuth token expiration after ~60 minutes. The silent refresh sends outdated OAuth v1 resource format instead of v2 scope format, causing Azure AD rejection (AADSTS90009). Long-running coding sessions die unexpectedly without re-auth flow. Includes evidence for GitHub Copilot troubleshooting demand.
MCP silent token refresh sends v1 resource=<clientId> instead of v2 scope= → AADSTS90009 after ~60 min / GitHub Copilot API token expires mid-session with no proactive refresh, causing 401 errors in long-running subagents CI/CD pipelines fail with AADSTS7000222 when Azure AD service principal secrets expire; developers need to rotate credentials without breaking builds. Includes evidence for Azure troubleshooting demand.
AADSTS7000222: The provided client secret keys for app are expired — Azure DevOps pipeline fails to authenticate to Azure Resource Manager Deployments from Google Cloud Build to Cloud Run fail with 403 Permission Denied when using custom service accounts; the roles required extend beyond just Cloud Build and Cloud Run. Includes evidence for GCP troubleshooting demand.
403 Permission Denied on Cloud Run Deploy / Cloud Build custom service account — even with Owner role Azure Pipelines jobs that run longer than expected (e.g., compilation, testing suites) silently lose authentication mid-execution and fail at the finalization stage, requiring investigation of agent timeout and credential refresh settings. Includes evidence for Azure troubleshooting demand.
[BUG]: Long-running jobs lose authentication and fail at finalization — Azure Pipelines agent loses session token during extended execution Fix Claude Code MCP OAuth flow where token acquisition appears successful but the server continues requesting authentication, requiring manual restart or re-login Includes evidence for Claude Code troubleshooting demand.
Claude Code OAuth authentication completes but MCP server stays in 'Needs authentication' state; Bearer token never sent to downstream server Fix Claude Code MCP connection/auth failures caused by OAuth timeout, locked keychain, or DNS preventing auth endpoint reachability Includes evidence for Claude Code troubleshooting demand.
Authorization with the MCP server failed / Couldn't reach the MCP server — OAuth callback timeouts, macOS keychain locked, or DNS resolution blocking the Anthropic auth endpoint Fix Docker Compose 'port is already allocated' error when deploying multi-replica services; learn proper port mapping for replicated containers using service discovery instead of direct port binding Includes evidence for Docker troubleshooting demand.
port is already allocated — docker-compose fails when using replicas with explicit port mapping; listen tcp 0.0.0.0:PORT: bind: address already in use Diagnose why Docker reserves port internally even after stopping/removing container; find cause of stale port reservation that persists through docker-compose down and reboot Includes evidence for Docker troubleshooting demand.
bind: address already in use — netstat -a -b -o reveals no service listening on port but Docker still refuses to bind; occurs after Docker update or container kill/cleanup cycle Claude Code 用户遇到认证失败（OAuth token 过期或 API key 无效），需要在本地开发环境或无头服务器上重新配置身份验证以恢复编码能力。 Includes evidence for Claude Code troubleshooting demand.
Authentication failed. Run `claude auth login` to authenticate. / Error: 401 - {"error":{"type":"authentication_error","message":"invalid x-api-key"}} Cursor IDE 用户面临多种错误（API key 无效、模型不可用、网络超时），需要系统性的故障排除指南来恢复 AI 辅助编程功能。 Includes evidence for Cursor troubleshooting demand.
Invalid API Key / Model not available or Unsupported model / Request timeout / Connection failed / Network error Fix GitHub Actions workflows failing due to default GITHUB_TOKEN having only read-only access, preventing PR creation, branch pushes, or script execution Includes evidence for GitHub Actions troubleshooting demand.
github-actions[bot] denied permission / GITHUB_TOKEN lacks repository write access Fix recurring elevated error rates on Claude Opus 4.7 — model frequently returns errors despite being the premium tier Includes evidence for Anthropic API troubleshooting demand.
Elevated errors on Claude Opus 4.7 model requests Fix Claude Code MCP OAuth authentication failure where no browser opens and no tools appear despite correct configuration Includes evidence for Claude Code troubleshooting demand.
HTTP MCP servers requiring OAuth silently fail with no user notification — tools/list returns Unauthorized but Claude Code shows nothing; debug log reveals 'Redirection handling is disabled, skipping redirect' Fix Docker EACCES permission denied errors when installing global npm packages or running build scripts inside containers Includes evidence for Docker troubleshooting demand.
EACCES: permission denied — mkdir /usr/local/lib/node_modules/.npm/_cacache/tmp or node_modules during npm install global package in Docker/container Debug LiteLLM proxy when routed API calls time out or models become unavailable across integrated providers (OpenAI, Anthropic, local models) Includes evidence for LiteLLM troubleshooting demand.
LiteLLM proxy timeout — model unreachable or provider timeout on multi-provider routing fallback Distinguish Anthropic HTTP 529 overloaded_error (server capacity saturation) from 429 rate limit errors; implement correct retry strategy for each Includes evidence for Anthropic API troubleshooting demand.
anthropic.APIStatusError: status_code=529, anthropic.error.overloaded_error: The API is temporarily overloaded. Paying Pro+ users hit weekly rate limit despite subscription; auto mode sometimes works but manual model selection instantly triggers rate limit error Includes evidence for GitHub Copilot troubleshooting demand.
"You've reached your weekly rate limit. Please upgrade your plan or wait for your limit to reset on <date>" / "Failed to connect to API: 429 Too Many Requests user_global_rate_limited:pro_plus" Fix Cursor IDE crashing repeatedly (code 5) after extended agent/composer sessions, caused by memory leak or SQLite nested transaction bug Includes evidence for Cursor troubleshooting demand.
Cursor crashed (code 5) | Renderer OOM when agent turn loads many file contexts + large writes | RangeError: Invalid array length in setCardContentReady crashes renderer (code 5) Fix persistent Copilot Pro+ weekly rate limit that does not reset, blocking paid subscribers from using any model Includes evidence for GitHub Copilot troubleshooting demand.
Sorry, you've exceeded your 5 hour session limits. Please review our Terms of Service. Code: user_weekly_rate_limited:pro_plus Debug Kubernetes CrashLoopBackOff pod with empty logs, distinguish between application crashes vs node resource eviction (OOM/memory pressure) Includes evidence for Kubernetes troubleshooting demand.
CrashLoopBackOff | kubelet eviction node memory pressure/disk pressure | OOMKilled (Exit Code 137) Understand and mitigate LiteLLM Proxy auth bypass via SQL injection where DB connection errors grant PROXY_ADMIN privileges without valid keys Includes evidence for LiteLLM troubleshooting demand.
Authentication failed / failed-to-connect-to-db / PrismaError pool timeout connection error Fix Anthropic API 529 overloaded errors during peak traffic or model launches; implement retry-and-fallback strategy for production apps Includes evidence for Anthropic API troubleshooting demand.
HTTP 529 overloaded_error: "API Error: 529 overloaded" — global capacity saturation, not per-key rate limit Fix actions/checkout action failing with two new error messages after workflow permission changes restrict GITHUB_TOKEN contents scope to none Includes evidence for GitHub Actions troubleshooting demand.
repository 'https://github.com/...' not found OR could not read Username for 'https://github.com': terminal prompts disabled — actions/checkout GITHUB_TOKEN contents scope set to none Fix Lambda functions that timeout during the Init phase before Invoke can run; diagnose cold start delays, initialization code bottlenecks, or dependency loading issues Includes evidence for AWS troubleshooting demand.
Task timed out after 3.00 seconds — Lambda function times out during Init phase (Sandbox.Timedout) Diagnose and fix pods stuck in CrashLoopBackOff specifically caused by memory limits being exceeded (OOMKilled); determine whether to increase resources or fix memory leaks Includes evidence for Kubernetes troubleshooting demand.
CrashLoopBackOff with Reason: OOMKilled, Exit Code 137 Determine if a project installed compromised LiteLLM versions and clean the installation; scan GitLab CI/CD jobs and GitHub Actions for affected deployments Includes evidence for LiteLLM troubleshooting demand.
LiteLLM package compromised — versions 1.82.7 and 1.82.8 contain malicious code contacting external C2 domains Fix ollama pulling manifest failures when importing custom GGUF models from HuggingFace; resolve sha256 blob loading errors after successful download Includes evidence for Ollama troubleshooting demand.
pulling manifest Error: 500 Internal Server Error: unable to load model: C:\Users\...ollama\models\blobs\sha256-... Fix Next.js/Vercel deploy module not found errors caused by case sensitivity differences between local Windows/macOS filesystem and Vercel's Linux build environment Includes evidence for Vercel troubleshooting demand.
Module not found: Can't resolve '@/components/forms/RegisterForm' — Build successful locally but fails on Vercel deployment Diagnose and fix Cloudflare 520 error when origin server returns garbled or empty response after TCP connection succeeds, distinguishing from 522 timeout and 524 gateway timeout Includes evidence for Cloudflare troubleshooting demand.
"Error 520: web server returns an unknown error" — connection established but response is empty/malformed/reset