• API
• pg.Pool

new Pool

new Pool(config: Config)

Constructs a new pool instance.

The pool is initially created empty and will create new clients lazily as they are needed. Every field of the config object is entirely optional. The config passed to the pool is also passed to every client instance within the pool when the pool creates that client.

type Config = {
  // all valid client config options are also valid here
  // in addition here are the pool specific configuration parameters:
 
  // number of milliseconds to wait before timing out when connecting a new client
  // by default this is 0 which means no timeout
  connectionTimeoutMillis?: number
 
  // number of milliseconds a client must sit idle in the pool and not be checked out
  // before it is disconnected from the backend and discarded
  // default is 10000 (10 seconds) - set to 0 to disable auto-disconnection of idle clients
  idleTimeoutMillis?: number
 
  // maximum number of clients the pool should contain
  // by default this is set to 10.
  max?: number
 
  // Default behavior is the pool will keep clients open & connected to the backend
  // until idleTimeoutMillis expire for each client and node will maintain a ref
  // to the socket on the client, keeping the event loop alive until all clients are closed
  // after being idle or the pool is manually shutdown with `pool.end()`.
  //
  // Setting `allowExitOnIdle: true` in the config will allow the node event loop to exit
  // as soon as all clients in the pool are idle, even if their socket is still open
  // to the postgres server.  This can be handy in scripts & tests
  // where you don't want to wait for your clients to go idle before your process exits.
  allowExitOnIdle?: boolean
}

example to create a new pool with configuration:

import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool({
  host: 'localhost',
  user: 'database-user',
  max: 20,
  idleTimeoutMillis: 30000,
  connectionTimeoutMillis: 2000,
})

pool.query

Often we only need to run a single query on the database, so as convenience the pool has a method to run a query on the first available idle client and return its result.

pool.query(text: string, values?: any[]) => Promise<pg.Result>

import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool()
 
const result = await pool.query('SELECT $1::text as name', ['brianc'])
console.log(result.rows[0].name) // brianc

Notice in the example above there is no need to check out or release a client. The pool is doing the acquiring and releasing internally. I find pool.query to be a handy shortcut many situations and use it exclusively unless I need a transaction.

pool.connect

pool.connect() => Promise<pg.Client>

Acquires a client from the pool.

• If there are idle clients in the pool one will be returned to the callback on process.nextTick.
• If the pool is not full but all current clients are checked out a new client will be created & returned to this callback.
• If the pool is 'full' and all clients are currently checked out will wait in a FIFO queue until a client becomes available by it being released back to the pool.

import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool()
 
const client = await pool.connect()
await client.query('SELECT NOW()')
client.release()

releasing clients

client.release(destroy?: boolean) => void

Client instances returned from pool.connect will have a release method which will release them from the pool.

The release method on an acquired client returns it back to the pool. If you pass a truthy value in the destroy parameter, instead of releasing the client to the pool, the pool will be instructed to disconnect and destroy this client, leaving a space within itself for a new client.

import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool()
 
// check out a single client
const client = await pool.connect()
 
// release the client
client.release()

import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool()
assert(pool.totalCount === 0)
assert(pool.idleCount === 0)
 
const client = await pool.connect()
await client.query('SELECT NOW()')
assert(pool.totalCount === 1)
assert(pool.idleCount === 0)
 
// tell the pool to destroy this client
await client.release(true)
assert(pool.idleCount === 0)
assert(pool.totalCount === 0)

pool.end

Calling pool.end will drain the pool of all active clients, disconnect them, and shut down any internal timers in the pool. It is common to call this at the end of a script using the pool or when your process is attempting to shut down cleanly.

// again both promises and callbacks are supported:
import pg from 'pg'
const { Pool } = pg
 
const pool = new Pool()
 
await pool.end()

properties

pool.totalCount: number

The total number of clients existing within the pool.

pool.idleCount: number

The number of clients which are not checked out but are currently idle in the pool.

pool.waitingCount: number

The number of queued requests waiting on a client when all clients are checked out. It can be helpful to monitor this number to see if you need to adjust the size of the pool.

events

Pool instances are also instances of EventEmitter.

connect

pool.on('connect', (client: Client) => void) => void

Whenever the pool establishes a new client connection to the PostgreSQL backend it will emit the connect event with the newly connected client. This presents an opportunity for you to run setup commands on a client.

const pool = new Pool()
pool.on('connect', (client) => {
  client.query('SET DATESTYLE = iso, mdy')
})

acquire

pool.on('acquire', (client: Client) => void) => void

Whenever a client is checked out from the pool the pool will emit the acquire event with the client that was acquired.

error

pool.on('error', (err: Error, client: Client) => void) => void

When a client is sitting idly in the pool it can still emit errors because it is connected to a live backend.

If the backend goes down or a network partition is encountered all the idle, connected clients in your application will emit an error through the pool's error event emitter.

The error listener is passed the error as the first argument and the client upon which the error occurred as the 2nd argument. The client will be automatically terminated and removed from the pool, it is only passed to the error handler in case you want to inspect it.

release

pool.on('release', (err: Error, client: Client) => void) => void

Whenever a client is released back into the pool, the pool will emit the release event.

remove

pool.on('remove', (client: Client) => void) => void

Whenever a client is closed & removed from the pool the pool will emit the remove event.