Windows 11 local source build troubleshooting

Use WSL2 for the source build and keep your browser, GitHub OAuth App, and local environment aligned to the same URL scheme. Most Windows-specific failures come from a mismatch between the Vite dev server, the Go backend, and the loopback address used during OAuth.

Before you start

• Run ./start-dev.sh or ./startup-oauth.sh inside WSL2.
• Prefer localhost everywhere unless you intentionally switch to 127.0.0.1.
• Verify which mode you are using before debugging ports:

Command	Browser entrypoint	Backend target	OAuth required
./start-dev.sh	http://localhost:5174	http://localhost:8080	No
./startup-oauth.sh	http://localhost:8080	http://localhost:8081 behind the watchdog on :8080	Yes
./startup-oauth.sh --dev	http://localhost:5174	http://localhost:8081 via the Vite proxy and watchdog	Yes

Port configuration: Vite on :5174, backend on :8080 or :8081

Symptom: the UI opens, but API calls fail, the login button spins forever, or the browser shows 404, 502, or WebSocket errors.

Cause: on source builds, the browser usually talks to Vite on :5174, while the API still lives on the Go backend. The checked-in Vite config proxies /api, /auth/github, /auth/github/callback, and /ws to the backend port.

const backendPort = process.env.BACKEND_LISTEN_PORT || '8081'
const target = `http://localhost:${backendPort}`
 
proxy: {
  '/api': { target, changeOrigin: true },
  '/auth/github': { target, changeOrigin: true },
  '/auth/github/callback': { target, changeOrigin: true },
  '/ws': { target: `ws://localhost:${backendPort}`, ws: true },
}

What to use on Windows

• For ./start-dev.sh, keep the browser on http://localhost:5174 and make sure the proxy targets port 8080.
• For ./startup-oauth.sh --dev, keep the browser on http://localhost:5174, but let the proxy target the backend behind the watchdog on port 8081.
• For ./startup-oauth.sh without --dev, do not use Vite at all; open http://localhost:8080.

If you run Vite manually, export the backend port explicitly before starting it:

# start-dev.sh style
export BACKEND_LISTEN_PORT=8080
npm run dev -- --port 5174

Loopback mismatches: localhost vs 127.0.0.1

Symptom: GitHub redirects back with redirect_uri_mismatch, csrf_validation_failed, or the callback lands on the wrong page.

Cause: on Windows, localhost and 127.0.0.1 are usually equivalent for humans, but GitHub OAuth treats them as different redirect URIs. Mixing them across WSL, the Windows browser, the GitHub OAuth App, and the local script setup breaks the handshake.

Recommended fix

Pick one hostname and use it everywhere. The easiest path is:

• Browser URL: http://localhost:5174 or http://localhost:8080
• GitHub OAuth App homepage URL: http://localhost:5174 for --dev, otherwise http://localhost:8080
• GitHub OAuth callback URL: http://localhost:8080/auth/github/callback
• Any manual overrides such as FRONTEND_URL: use localhost, not 127.0.0.1

If you decide to use 127.0.0.1, update all of those values together and clear cookies for the old host before retrying.

Environment sync: GitHub App settings and .env must match

Symptom: the sign-in page loads, but GitHub rejects the callback or the backend cannot complete the token exchange.

Create .env in the repo root (the same directory as startup-oauth.sh) and then align it with the GitHub OAuth App.

GITHUB_CLIENT_ID=your_client_id
GITHUB_CLIENT_SECRET=your_client_secret

Checklist

1. Create the GitHub OAuth App in GitHub Developer Settings.
2. Copy the Client ID and Client Secret into .env.
3. Make the GitHub App URLs match the mode you actually run:

Mode	Homepage URL	Callback URL
./startup-oauth.sh	http://localhost:8080	http://localhost:8080/auth/github/callback
./startup-oauth.sh --dev	http://localhost:5174	http://localhost:8080/auth/github/callback

4. Restart the console after every .env or GitHub App change.
5. If you changed hostnames, remove cookies for the old local origin and sign in again.

./start-dev.sh does not require GitHub OAuth. Use it when you need a local dev-user session.

Missing ~/.kube/config in WSL2 or Windows

Symptom: cluster views stay empty, backend logs keep warning about kubeconfig, or kubectl-backed features fail immediately.

Cause: the console checks KUBECONFIG first and otherwise falls back to ~/.kube/config. On Windows 11, your kubeconfig may exist in the Windows profile, not inside WSL.

Fix option 1: copy the Windows kubeconfig into WSL

mkdir -p ~/.kube
cp /mnt/c/Users/<windows-user>/.kube/config ~/.kube/config
chmod 600 ~/.kube/config
kubectl config get-contexts

Fix option 2: point KUBECONFIG at the Windows file

export KUBECONFIG=/mnt/c/Users/<windows-user>/.kube/config
kubectl config get-contexts

Fix option 3: no-cluster / UI-only work

If you are working on the frontend, use demo data instead of a live kubeconfig:

cd web
VITE_DEMO_MODE=true npm run dev -- --port 5174

That gives you a clean UI-only workflow while you sort out kubeconfig access.

Quick verification commands

Run these from WSL2 after changes:

# Frontend reachable
curl -I http://localhost:5174
 
# Backend / watchdog reachable
curl -I http://localhost:8080
 
# kubeconfig visible to kubectl
kubectl config get-contexts

From PowerShell, use curl.exe or Invoke-RestMethod instead of the PowerShell curl alias.