Connection pooling with PgBouncer in Supabase

Introduction

Connection pooling helps manage many database requests efficiently by reusing a few connections. PgBouncer is a tool that does this for PostgreSQL databases.

When your app has many users connecting to the database at the same time.

When you want to reduce the time it takes to open new database connections.

When your database has a limit on how many connections it can handle.

When you want to improve app performance by managing database load better.

Syntax

Supabase

[databases]
mydb = host=localhost port=5432 dbname=mydb user=myuser password=mypass

[pgbouncer]
listen_addr = *
listen_port = 6432
auth_type = md5
auth_file = /etc/pgbouncer/userlist.txt
pool_mode = session
max_client_conn = 100
default_pool_size = 20

The [databases] section defines which databases PgBouncer will connect to.

The [pgbouncer] section sets how PgBouncer listens and manages connections.

Examples

Basic setup using session pooling mode, where each client connection gets a dedicated server connection during the session.

Supabase

[databases]
mydb = host=localhost port=5432 dbname=mydb user=myuser password=mypass

[pgbouncer]
pool_mode = session

Transaction pooling mode, where server connections are assigned only during a transaction, improving efficiency for short queries.

Supabase

[databases]
mydb = host=localhost port=5432 dbname=mydb user=myuser password=mypass

[pgbouncer]
pool_mode = transaction

Statement pooling mode, where each statement uses a server connection, best for very short queries but with some limitations.

Supabase

[databases]
mydb = host=localhost port=5432 dbname=mydb user=myuser password=mypass

[pgbouncer]
pool_mode = statement

Sample Program

This is a complete PgBouncer configuration for a Supabase PostgreSQL database. It uses transaction pooling to efficiently manage connections. It listens on all network interfaces on port 6432 and logs connections and disconnections.

Supabase

[databases]
supabase = host=db.supabase.co port=5432 dbname=supabase user=supabase_user password=secret_password

[pgbouncer]
listen_addr = 0.0.0.0
listen_port = 6432
auth_type = md5
auth_file = /etc/pgbouncer/userlist.txt
pool_mode = transaction
max_client_conn = 100
default_pool_size = 20
log_connections = 1
log_disconnections = 1

Important Notes

Always secure your PgBouncer auth_file with proper permissions to protect passwords.

Choose pool_mode based on your app's query patterns for best performance.

Monitor PgBouncer logs to understand connection usage and troubleshoot issues.

Summary

PgBouncer helps reuse database connections to improve app speed and reduce load.

Configure PgBouncer with your database details and choose a pooling mode.

Use transaction pooling for most apps to balance performance and compatibility.

Practice

1. What is the main purpose of using PgBouncer with a Supabase database?

easy

A. To increase the size of the database storage

B. To encrypt data between the app and database

C. To backup the database automatically

D. To reuse database connections and reduce load on the database

Solution

1. Step 1: Understand PgBouncer's role

   PgBouncer is a tool that manages and reuses database connections to avoid opening new ones repeatedly.

2. Step 2: Identify the benefit

   Reusing connections reduces the load on the database and improves app speed by avoiding connection overhead.

3. Final Answer:

   To reuse database connections and reduce load on the database -> Option D

4. Quick Check:

   Connection reuse = reduce load [OK]

Hint: PgBouncer saves time by reusing connections [OK]

Common Mistakes:

• Thinking PgBouncer encrypts data
• Confusing PgBouncer with backup tools
• Assuming it increases storage size

2. Which of the following is the correct way to specify the pooling mode in PgBouncer's configuration file?

easy

A. pool_mode = transaction

B. pooling-mode: session

C. poolMode = statement

D. pool_mode = connect

Solution

1. Step 1: Recall PgBouncer config syntax

   PgBouncer uses key = value format, with underscores in keys.

2. Step 2: Identify correct pooling mode syntax

   The correct key is pool_mode and valid values include transaction, session, and statement.

3. Final Answer:

   pool_mode = transaction -> Option A

4. Quick Check:

   Correct key and value format = pool_mode = transaction [OK]

Hint: Look for key = value with underscores [OK]

Common Mistakes:

• Using hyphens instead of underscores
• Capitalizing keys incorrectly
• Using invalid pooling mode names

3. Given this PgBouncer config snippet:

pool_mode = transaction
max_client_conn = 100
default_pool_size = 20

What happens when 50 clients connect simultaneously and each runs a short transaction?

medium

A. PgBouncer queues all clients until connections free up, causing delays

B. PgBouncer rejects connections after 20 clients because of pool size

C. PgBouncer allows all 50 clients to share 20 server connections efficiently

D. Each client gets a dedicated server connection, so 50 connections open

Solution

1. Step 1: Understand transaction pooling mode

   In transaction mode, PgBouncer assigns server connections only during active transactions, allowing reuse.

2. Step 2: Analyze connection limits

   default_pool_size = 20 means max 20 server connections per database/user pool, but clients can exceed this by sharing connections.

3. Final Answer:

   PgBouncer allows all 50 clients to share 20 server connections efficiently -> Option C

4. Quick Check:

   Transaction pooling shares connections = PgBouncer allows all 50 clients to share 20 server connections efficiently [OK]

Hint: Transaction pooling shares connections among clients [OK]

Common Mistakes:

• Assuming one client = one server connection
• Thinking pool size limits client connections
• Confusing session pooling with transaction pooling

4. You configured PgBouncer with pool_mode = transaction but your app experiences errors with prepared statements. What is the likely cause?

medium

A. PgBouncer does not support prepared statements in any mode

B. Transaction pooling reuses connections between clients, causing prepared statement conflicts

C. Transaction pooling is required for prepared statements to work

D. The max_client_conn setting is too low

Solution

1. Step 1: Understand transaction pooling behavior

   In transaction mode, PgBouncer assigns server connections only during transactions and reuses them for other clients after reset.

2. Step 2: Identify cause of errors

   Prepared statements fail because the connection state (including prepares) is reset (DISCARD ALL) between transactions/clients.

3. Final Answer:

   Transaction pooling reuses connections between clients, causing prepared statement conflicts -> Option B

4. Quick Check:

   Transaction pooling = resets between transactions [OK]

Hint: Transaction pooling reuses connections, watch prepared statement issues [OK]

Common Mistakes:

• Believing PgBouncer supports prepared statements in transaction mode
• Confusing transaction and session pooling effects
• Blaming max_client_conn for statement errors

5. You want to optimize a Supabase app with many short queries and avoid connection overhead. Which PgBouncer pooling mode should you choose and why?

hard

A. Transaction pooling, because it reuses connections only during transactions for best performance

B. Session pooling, because it keeps connections open for each client session

C. Statement pooling, because it assigns a connection per SQL statement

D. No pooling, to avoid any connection sharing issues

Solution

1. Step 1: Consider app query pattern

   Many short queries mean frequent connections and disconnections, so reusing connections efficiently is key.

2. Step 2: Evaluate pooling modes

   Transaction pooling assigns connections only during transactions, allowing reuse and reducing overhead without tying connections to sessions.

3. Step 3: Compare other modes

   Session pooling holds connections longer, increasing resource use; statement pooling is too strict and can cause errors; no pooling wastes resources.

4. Final Answer:

   Transaction pooling, because it reuses connections only during transactions for best performance -> Option A

5. Quick Check:

   Short queries + reuse = transaction pooling [OK]

Hint: Use transaction pooling for many short queries [OK]

Common Mistakes:

• Choosing session pooling for short queries
• Thinking statement pooling is best for performance
• Avoiding pooling altogether