Postgres connection pooling with pg + max lifetime

12498 views

                  import { Pool } from 'pg';

export const pool = new Pool({
  connectionString: process.env.DATABASE_URL,
  max: Number(process.env.PG_POOL_MAX ?? 10),
  idleTimeoutMillis: 30_000,
  connectionTimeoutMillis: 3_000,
  application_name: process.env.SERVICE_NAME ?? 'codesnips-api'
} as any);

export async function withClient<T>(fn: (client: import('pg').PoolClient) => Promise<T>): Promise<T> {
  const client = await pool.connect();
  try {
    return await fn(client);
  } finally {
    client.release();
  }
}

After getting burned by long-lived connections that slowly accumulate bad state (or get killed by the network) and then explode during peak traffic, I got strict about pg pooling. I keep the pool size small per instance and scale horizontally instead of cranking max and hoping for the best. I also set application_name so DBA tooling can immediately attribute connections to the right service. The small but critical detail is a connection acquisition timeout (e.g. connectionTimeoutMillis): without it, requests can hang and you’ll misdiagnose it as ‘the API is slow’ when the real issue is ‘the DB pool is exhausted’. Once pool timeouts are first-class errors, capacity planning becomes much clearer.

Mateo Rodriguez

More from Mateo Rodriguez

Postgres connection pooling with pg + max lifetime

0 Comments

More from Mateo Rodriguez