Three Reasons Your Lovable App Stopped Working (And How to Fix Each One)
Your Lovable app was working fine. Now it's not. Login fails, pages go blank, users are complaining, and you're not sure if you broke something or the internet is broken.
Before you start debugging your code, take a breath. It's almost always one of three things - and none of them are your fault.
The three usual suspects
| Symptom | Likely cause | Jump to |
|---|---|---|
| "Fail to Fetch" on login, auth errors across all projects | Supabase is having issues | Supabase issues |
| 503/522 errors, slow responses, timeouts | Your instance is overloaded | Instance overload |
| Blank/white page, "supabaseUrl is required" in console | Build didn't inject env vars | Blank pages |
Step 0: Check the status pages
Before anything else, check these two links:
- Supabase Status - real-time status of Auth, API Gateway, Database, Storage, Edge Functions
- Lovable Status - Lovable editor and platform status
If you see active incidents or partial outages, that's your answer. Stop debugging your code.
Supabase issues
Lovable uses Supabase for auth, database, storage, realtime, and edge functions. When Supabase has issues, your app feels broken even though your code is perfectly fine.
How to tell: multiple projects affected at once, "Fail to Fetch" errors on login, Supabase status page shows incidents.
Quick fixes to get yourself working
These get you back to building. They won't fix it for your end users, but they'll get you unstuck.
Cloudflare WARP (free, fastest)
Download from 1.1.1.1, install, toggle on, try your app again. WARP reroutes your traffic through Cloudflare's network, which can bypass regional routing issues. No account needed.
VPN
Connect to a server in a different region. If the issue is regional, routing through an unaffected area gets you back in immediately.
Change your DNS
Sometimes the issue is at the DNS level. Switching to a public DNS can help:
| DNS Provider | Primary | Secondary | Notes |
|---|---|---|---|
| Cloudflare | 1.1.1.1 | 1.0.0.1 | Fastest, privacy-focused |
| 8.8.8.8 | 8.8.4.4 | Reliable, widely used | |
| Quad9 | 9.9.9.9 | 149.112.112.112 | Security-focused |
How to change it:
- Windows: Settings > Network & Internet > Wi-Fi > your network > Edit DNS
- Mac: System Settings > Wi-Fi > Details on your network > DNS
- Phone: Wi-Fi settings > tap your network > DNS
Close your browser completely and reopen after changing.
Fixing it for your end users
If your app has real users who are affected, you need a deeper fix.
Own Supabase project - Custom Domain (best option)
Instead of your app calling yourproject.supabase.co, it calls something like api.yourdomain.com, which Cloudflare proxies to Supabase. This works with all auth methods including OAuth.
What you need:
- Supabase Pro plan ($25/month)
- A domain on Cloudflare
- Custom domain add-on ($10/month)
Steps:
- Move your domain to Cloudflare if it's not there yet
- In Supabase dashboard > Settings > Custom Domains, set up your domain
- Make sure the DNS record in Cloudflare is set to Proxied (orange cloud ON)
- Update your app's SUPABASE_URL to the new custom domain
- Test auth, database, and realtime
Own Supabase project - Cloudflare Worker proxy (free)
If you're on the free plan, set up a Cloudflare Worker as a reverse proxy. Free tier gives you 100,000 requests/day. Works for email/password auth, but OAuth won't work through it.
Lovable Cloud - consider migrating
If outages keep affecting your users and you need more control, migrating to your own Supabase project gives you full control over your database, environment, and configuration - while keeping the Lovable editor experience. Migration guide here.
Regional blocks (ISP-level)
Sometimes the issue isn't a Supabase outage but an ISP blocking access to Supabase domains. We've seen this in India (resolved March 2026), Argentina (Movistar), and Brazil (Claro). Same symptoms, different root cause. If the status pages look fine but your region is affected, the WARP/VPN fixes above still work. For the full deep dive on regional blocks: Supabase Blocked? Fix Your Lovable App.
A note on security
Don't route your traffic through third-party proxy services you find online. Your API keys, user passwords, and auth tokens all pass through those proxies. If you don't control the proxy, you don't control who sees your data. Stick to your own Cloudflare Worker or Supabase custom domain.
Instance overload
Your app works sometimes but throws 503 or 522 errors, especially under load. Responses are slow, timeouts happen randomly.
How to tell: only one project is affected (not all of them), Supabase status page looks fine, errors happen more during peak usage.
Check your instance tier
Go to your Lovable project Settings > Cloud & AI balance. Check what instance tier you're on. If you're on the smallest tier and your app has real users, you probably need to upgrade.
Instance size controls your database resources (CPU, memory, connections). It doesn't affect edge function limits - those are platform-level.
The refresh trick
If your Lovable Cloud project gets stuck in a broken state - "Project has been removed" errors, edge functions failing, database inaccessible - try this:
- Downsize your instance tier (go one level down)
- Wait for the change to apply
- Upgrade it back to where it was (or higher)
This triggers a platform refresh on Lovable's side. It can take up to an hour, but it often fixes stuck instances. This was discovered by Marcos in the community and has worked for several users.
Deep dive: understanding instance sizes
For a full breakdown of instance tiers (Tiny through Large), resource alerts to watch for (disk, IO budget, CPU), and when to upgrade, check out: Lovable Cloud 503 Errors - What's Actually Happening.
When to email support
If the 503/522 errors persist after upgrading and refreshing, email support@lovable.dev with your project ID. Some infrastructure issues can only be resolved on their end.
Blank pages
Your app loads but shows a completely white/blank screen. No content, no error visible to the user.
How to tell: open F12 > Console and look for red errors.
"supabaseUrl is required"
This means Lovable's build didn't inject the VITE_SUPABASE_URL and VITE_SUPABASE_PUBLISHABLE_KEY environment variables. This can happen after a mid-build credit interruption, a failed publish, or seemingly at random.
Fix: add hardcoded fallbacks in your Supabase client file. Paste this prompt in Lovable:
"In my supabase client file (src/integrations/supabase/client.ts or src/lib/supabase.ts), add hardcoded fallbacks for the env vars so the app doesn't break if they aren't injected. Use my project URL and anon key as fallbacks like this: const supabaseUrl = import.meta.env.VITE_SUPABASE_URL || 'https://[my-project-ref].supabase.co' and do the same for the anon key. Don't remove the env var references, just add the || fallback."
You can find your project URL and anon key in the Cloud panel. Anon keys are public by design, so hardcoding them as a fallback is safe.
Blank screen after Google login
If the screen goes blank specifically after a Google OAuth login, the issue is Supabase's implicit auth flow putting tokens in the URL hash, which conflicts with client-side routing.
Fix: paste this prompt in Lovable:
"Add a clearSupabaseAuthHash function that runs on app startup. It should check if window.location.hash contains access_token, refresh_token, or token_type. If it does, use window.history.replaceState to clear the hash while keeping the pathname and search params."
Published site shows old code
If your preview works fine but the published site looks wrong or shows old content:
- Make a small change (add a comment or space)
- Publish again
- If it's still stale, remix the project and re-publish
The published site uses a CDN that can cache aggressively. A fresh publish usually fixes it.
Quick reference
| Problem | First thing to check | Quick fix |
|---|---|---|
| All projects down | status.supabase.com | Cloudflare WARP or VPN |
| One project, 503 errors | Instance tier | Upgrade or refresh instance |
| Blank page, console error | F12 > Console | Env var fallback prompt |
| Blank after Google login | URL has #access_token | Auth hash clear prompt |
| Published site stale | Preview vs published | Remix and re-publish |
| Can't login to Lovable itself | status.lovable.dev | Wait, or try different browser |
Useful links
Reader margin
What stayed with you?
Leave one small mark in the margin.
Notes from readers
The margin is open. You can leave the first note.
Stuck with something like this? I fix it daily → Services