Back to articles

Your Supabase App Broke in India. It Wasn't Your Code.

Feb 28, 2026Dishant Sharma7 min read
Your Supabase App Broke in India. It Wasn't Your Code.

February 24, 2026. Developers across India opened their laptops and found their apps broken. Not a bug. Not a deployment failure. Jio, India's largest telecom with 500 million subscribers, had silently blocked Supabase.

No announcement. No warning. No migration window.

One developer on Reddit summed up the mood in five words: "can't have shit in india." That comment got 92 upvotes. That's not anger. That's exhaustion.

If you've built anything with Supabase, you know the specific panic of production going dark. Auth stops responding. The database returns nothing. You start checking your schema, rereading your connection string, redeploying. And nothing changes. Because nothing in your code is wrong.

Supabase's servers were running fine. The infrastructure was healthy. The problem was between your users and those servers. And that's what makes this so hard to stomach.

India is Supabase's fourth-largest market. Indian visits jumped 179% year over year, hitting roughly 365,000 in January 2026 alone. Thousands of indie developers, student projects, early-stage startups. All on a platform with a generous free tier that didn't ask you to be a DevOps expert.

And overnight, it was unreachable.

You've probably hit a production incident before. i know i have. But this one is different. You can't hotfix a government order.


What actually got blocked

I used to think a "block" meant the website wouldn't load. That's not what happened here.

Supabase's main website loaded for most people. But the underlying infrastructure was blocked. The API endpoints, the auth service, the database connections, all the supabase.co domains your app actually talks to.

That distinction matters. Users saw your app behaving strangely and assumed it was your fault. It wasn't. The ISPs were returning bad DNS responses for supabase.co domains. Jio first. Then Airtel. Then ACT Fibernet. Three of India's biggest providers, all enforcing the same block.

Your app looked broken. Your app was broken. But not because of your code.

Supabase's status page documented this in real time. At 14:46 UTC on February 24, they flagged connectivity complaints from the AP South 1 region. By 15:13 UTC, they'd identified the DNS resolution failure. By 16:00 UTC, they confirmed it was an ISP-level block inside India.


Why Section 69A is the real problem

Here's a question people keep asking: what law lets a government block developer infrastructure overnight, with zero explanation?

Section 69A of India's Information Technology Act. It gives the government broad power to restrict access to online content for national security, public order, or several other reasons.

It's the same law used to ban TikTok in 2020. The same law that blocked PUBG, WeChat, and hundreds of apps.

But those were consumer apps. Supabase is backend infrastructure. Blocking it doesn't remove content. It breaks someone else's product.

And the government is not required to explain why. No public statement was issued. Supabase found out exactly like developers did. Users reported broken connections, Supabase investigated, and the picture got clearer and worse as the days went on.

The Ministry of Electronics and Information Technology has said nothing publicly.


What Supabase did

The first time i saw a company publicly tag a government minister over a block, i assumed it would go nowhere.

Supabase tagged India's IT minister Ashwini Vaishnaw on X. They said millions of developers in India were affected. They asked for help.

Then they deleted the post.

No response came. Supabase kept updating the status page and told users to try three things:

  • Switch DNS to Cloudflare (1.1.1.1) or Google (8.8.8.8)

  • Use a VPN

  • Proxy Supabase through a custom domain

These work for a developer on their laptop. They don't work for your production users. Your users on Jio aren't changing their DNS settings because your login button stopped working.

A developer on Reddit put it plainly: "I'm not looking for client-side VPN/DNS fixes. I need infrastructure-level solutions."

That's the honest wall people hit.


The clever fix that isn't enough

Most tutorials tell you to change DNS and move on. Fine for local dev.

One developer built an open-source drop-in layer that automatically detects DNS failures and switches to DNS-over-HTTPS using Quad9 and Cloudflare. Available for Node.js, Python, Go, Rust, and Flutter. No VPN required.

It's genuinely clever. But it's a workaround for a problem that shouldn't exist.

The real infrastructure fix is proxying Supabase through your own custom domain using Cloudflare Workers or your own DNS layer. That way your app talks to your domain, not supabase.co. The block doesn't reach it.

i spent hours testing workarounds. The actual fix was a one-line config change. But figuring out which config took the hours.

And some developers have had no time at all. An entrepreneur told TechCrunch, anonymously, that they saw zero new user registrations from India for two to three days straight. Mid-launch. No users. No explanation. Just silence from the government.


A small thing about project names

Here's something unrelated. Sort of.

A lot of developers in India name their services after something personal. Sanskrit words. Planets. Random animals. i've seen PostgreSQL instances named "Shiva" and Redis caches named "Brahma."

There's something darkly funny about being on a call at 2 AM and someone says "Brahma is down." And you're not sure if they mean the cache or cosmological dissolution.

But the naming isn't random. It's ownership. When you name something, you feel responsible for it. You keep it alive. You notice when it breaks.

That's why this block feels personal. Supabase wasn't just a third-party service for a lot of Indian developers. It was the thing they built on. The free tier made real products possible for students who didn't have AWS budgets. Real products that are now unreachable for real users.

And nobody told them why.


The honest version

This block should make you rethink one thing: single points of failure.

Supabase is good. It's the right call for many projects. Low DevOps overhead, fast to prototype on, generous free tier. The love Indian developers have for it is not irrational.

But putting your entire production backend on one domain, accessible only if three governments and fifteen ISPs all cooperate, is fragile. Not because Supabase is bad. Because governments are unpredictable.

If you're building a side project or learning, keep using Supabase. None of this affects you meaningfully.

But if you're running a production app with real Indian users and no fallback plan, this is the moment. Custom domains. Proxy layers. Route traffic through your own domain.

And if you're in a regulated industry or serving enterprise clients, self-hosting Supabase was always the right call. It's open source. Run it on your own infrastructure and this block doesn't touch you.

Most people don't need that. But some people should have done it already and didn't.


This isn't over

That Reddit comment, "can't have shit in india," got 92 upvotes.

That's not a community complaining. That's a community that has seen this before. Medium was blocked. GitHub had issues. And now Supabase.

India has the fourth-biggest developer base on Supabase. Visits grew 179% in a year. The government blocked it anyway, with no warning, no reason, and no timeline for resolution.

Somewhere right now, there's a developer in Jaipur or Bangalore who spent three days assuming it was their bug. Checking their schema. Redeploying their Edge Functions. Running connection tests.

It wasn't their code.

It never was.

Recent posts

View all posts