How to fix P2024: Timed out fetching a new connection from the connection pool. Please consider reducing the number of requests or increasing the `connection_limit` parameter in Prisma

What this error means

When Prisma Client executes a query, it needs to acquire a database connection from its connection pool. The pool has a limited number of connections (default: `num_physical_cpus * 2 + 1`). When all connections are in use, new queries wait in a queue for up to 10 seconds (default pool timeout). If the query engine cannot obtain a connection within that time window—either because connections are held too long, there are too many concurrent queries, or the pool is too small—it throws a P2024 error and moves to the next query in the queue. This is particularly common in serverless environments (AWS Lambda, Vercel Functions, Cloudflare Workers) where each function instance creates its own connection pool, quickly exhausting the database's total connection limit. It also happens when using `Promise.all()` to run many Prisma queries in parallel without accounting for the connection limit.

Typical symptoms

• Queries fail intermittently during high traffic periods
• Error appears when using Promise.all() with many Prisma queries
• Serverless functions time out with P2024 error
• The error message includes 'Current limit: X' showing your pool size
• Queries succeed when running sequentially but fail when parallelized
• Application works fine locally but fails in production under load

Common causes

1. Too many concurrent queries relative to the connection pool size
2. Using Promise.all() to execute more Prisma queries than available connections
3. Serverless deployment with many concurrent function invocations sharing a database
4. Database connection limit is lower than total connections from all app instances
5. Long-running queries holding connections longer than expected
6. Not using an external connection pooler (PgBouncer, RDS Proxy, Supavisor) in serverless environments
7. Connection pool timeout (pool_timeout) is too short for your workload
8. Connection leaks preventing connections from being returned to the pool

How to fix "P2024: Timed out fetching a new connection from the connection pool. Please consider reducing the number of requests or increasing the `connection_limit` parameter"

# .env
DATABASE_URL="postgresql://user:password@host:5432/db?connection_limit=10"

# AWS Lambda / Vercel Functions - start with 1-3
DATABASE_URL="postgresql://user:password@host:5432/db?connection_limit=3"

# Traditional server with 8 CPUs: default is 17 (8*2+1)
# Increase to 25 if needed
DATABASE_URL="postgresql://user:password@host:5432/db?connection_limit=25"

# Default is 10 seconds, increase to 20
DATABASE_URL="postgresql://user:password@host:5432/db?pool_timeout=20"

DATABASE_URL="postgresql://user:password@host:5432/db?connection_limit=10&pool_timeout=15"

// If you have 50 items and connection_limit=10, this will fail
const results = await Promise.all(
  items.map(item => prisma.order.findUnique({ where: { id: item.id } }))
);

// Process in batches of 5
const chunkSize = 5;
const results = [];

for (let i = 0; i < items.length; i += chunkSize) {
  const chunk = items.slice(i, i + chunkSize);
  const chunkResults = await Promise.all(
    chunk.map(item => prisma.order.findUnique({ where: { id: item.id } }))
  );
  results.push(...chunkResults);
}

// One query uses one connection
const results = await prisma.order.findMany({
  where: { id: { in: items.map(i => i.id) } }
});

# Add pgbouncer=true to disable prepared statements
DATABASE_URL="postgresql://user:password@pgbouncer:6432/db?pgbouncer=true&connection_limit=1"

# Use Supabase's transaction pooler URL
DATABASE_URL="postgresql://user:password@pooler.supabase.com:6543/db?pgbouncer=true"

DATABASE_URL="postgresql://user:password@rds-proxy.region.rds.amazonaws.com:5432/db"

// Creates new client + pool every time
export async function handler(event) {
  const prisma = new PrismaClient();
  const data = await prisma.user.findMany();
  await prisma.$disconnect(); // DON'T do this in serverless
  return data;
}

// Reuse client across warm starts
const prisma = new PrismaClient({
  datasources: {
    db: { url: process.env.DATABASE_URL }
  }
});

export async function handler(event) {
  const data = await prisma.user.findMany();
  // Don't disconnect - reuse the connection
  return data;
}

aws lambda put-function-concurrency \
  --function-name my-function \
  --reserved-concurrent-executions 10

{
  "functions": {
    "api/**/*.ts": {
      "maxDuration": 10,
      "memory": 1024
    }
  }
}

Advanced notes

SELECT count(*) FROM pg_stat_activity WHERE datname = 'your_database';

SHOW max_connections;

const prisma = new PrismaClient({
  log: [
    { emit: 'event', level: 'query' },
  ],
});

prisma.$on('query', (e) => {
  console.log('Query: ' + e.query);
  console.log('Duration: ' + e.duration + 'ms');
});

Frequently asked questions

What does "P2024" mean?

When Prisma Client executes a query, it needs to acquire a database connection from its connection pool. The pool has a limited number of connections (default: num_physical_cpus * 2 + 1). When all connections are in use, new queries wait in a queue for up to 10 seconds (default pool timeout).

How do I fix "P2024"?

This error occurs when Prisma's connection pool is exhausted and cannot serve a query within the timeout period. It's most common in serverless environments, high-concurrency scenarios, or when running many parallel queries with insufficient connection limits.

What causes "P2024"?

Too many concurrent queries relative to the connection pool size Using Promise.all() to execute more Prisma queries than available connections Serverless deployment with many concurrent function invocations sharing a database

Is the "P2024" error serious?

Yes. This is a high-severity error that usually blocks work until it is fixed, but the steps above resolve it.

Related errors

• P2025: An operation failed because it depends on one or more records that were required but not found

  How to fix "P2025: An operation failed because it depends on one or more records that were required but not found" in Prisma

  MEDIUM

• P2023: Inconsistent column data

  How to fix "P2023: Inconsistent column data" in Prisma

  HIGH

• P2026: The current database provider doesn't support a feature that the query used

  How to fix "P2026: The current database provider doesn't support a feature that the query used" in Prisma

  MEDIUM

• P2022: The column `{column}` does not exist in the current database.

  The column does not exist in the current database

  HIGH

• P1002: The database server was reached but timed out

  Database server was reached but timed out

  HIGH

• P1013: The provided database string is invalid

  The provided database string is invalid

  HIGH

Official resources & further reading

• 📚Prisma Official Documentation
• 📖
  Connection pool - Prisma Documentation(Official Docs)
• 📖
  Database connections - Prisma Documentation(Official Docs)
• 📖
  Connection management - Prisma Guide(Official Docs)
• 📖
  Configure Prisma Client with PgBouncer - Prisma Documentation(Official Docs)
• 📖
  Deploy to AWS Lambda - Prisma Documentation(Official Docs)
• Overcoming Database Challenges in Serverless and Edge Applications - Prisma Blog
• 🐙
  Timed out fetching a new connection from the pool - GitHub Issue #7249(GitHub)
• 🐙
  Connection pool timeout - GitHub Discussion #20537(GitHub)
• 🐙
  Parallel execution with Promise.all causes P2024 error - GitHub Issue #22610(GitHub)
• Troubleshooting Prisma errors - Supabase Docs