I recently hit a frustrating issue while connecting Prisma to a Supabase Postgres database.
At first glance, everything looked fine:
- Correct credentials ✅
- Database status: healthy ✅
- Prisma schema: valid ✅
But then:
❌ "Can't reach database server at port 5432"
❌ "Network is unreachable"
⚠️ Prisma hangs when using connection pooler
🔍 What was actually happening?
The problem wasn’t Prisma.
It wasn’t Supabase either.
It was network-level IPv6 incompatibility.
My system resolved the database host to an IPv6 address, but my ISP/network doesn’t properly support IPv6 — so the connection failed silently.
⚙️ The breakdown
Supabase gives two connection types:
- Direct DB (port 5432)
- Requires stable IPv4/IPv6 support
- Used for migrations
- ❌ Failed in my case due to IPv6
- Connection Pooler (port 6543)
- Works over IPv4
- Designed for app runtime
- ⚠️ Not ideal for Prisma migrations
🛠️ Current workaround
- Using pooler for database connection
- Forcing Prisma to run with: PRISMA_CLIENT_ENGINE_TYPE=binary
- Skipping direct DB access for now
🤔 Open questions
- What’s the cleanest way to handle Prisma migrations without direct DB access?
- Are people relying on CI/CD runners or VPS environments for this?
- Is using a VPN the most practical fix long-term?
This was a good reminder that sometimes the issue isn’t your code or config — it’s the layer beneath everything.
Would love to hear how others have handled this.
Top comments (0)