Advisory locks in PostgreSQL - Mini Project: Build & Apply

Using Advisory Locks in PostgreSQL

📖 Scenario: You are managing a PostgreSQL database where multiple processes might try to update the same resource at the same time. To avoid conflicts, you want to use advisory locks to control access.

🎯 Goal: Learn how to create and use advisory locks in PostgreSQL to safely manage concurrent access to a shared resource.

📋 What You'll Learn

Create a table to simulate a shared resource

Define a lock key to use for advisory locking

Acquire an advisory lock before updating the resource

Release the advisory lock after the update

💡 Why This Matters

🌍 Real World

Advisory locks help prevent multiple processes from changing the same data at the same time, avoiding data corruption or conflicts.

💼 Career

Database administrators and backend developers use advisory locks to manage concurrency safely in multi-user environments.

1

Create a table to represent a shared resource

Create a table called shared_resource with two columns: id as an integer primary key and value as an integer. Insert one row with id = 1 and value = 100.

PostgreSQL

-- Create the shared_resource table and insert one row
-- Your code here

Hint

Use CREATE TABLE to define the table and INSERT INTO to add the initial row.

2

Define a lock key for advisory locking

Create a variable or note a constant integer lock_key with the value 12345 to use as the advisory lock key.

PostgreSQL

CREATE TABLE shared_resource (
    id INTEGER PRIMARY KEY,
    value INTEGER
);

INSERT INTO shared_resource (id, value) VALUES (1, 100);

-- Define the lock key
-- Your code here

Hint

Use the psql command \set to define a variable for the lock key.

3

Acquire the advisory lock before updating

Use the function pg_advisory_lock with the lock_key to acquire the advisory lock before updating the value in shared_resource. Increase the value by 10 where id = 1.

PostgreSQL

CREATE TABLE shared_resource (
    id INTEGER PRIMARY KEY,
    value INTEGER
);

INSERT INTO shared_resource (id, value) VALUES (1, 100);

-- Define the lock key
\set lock_key 12345

-- Acquire advisory lock and update value
-- Your code here

Hint

Use SELECT pg_advisory_lock(:lock_key) to lock, then update the table.

4

Release the advisory lock after update

Use the function pg_advisory_unlock with the lock_key to release the advisory lock after updating the shared_resource.

PostgreSQL

CREATE TABLE shared_resource (
    id INTEGER PRIMARY KEY,
    value INTEGER
);

INSERT INTO shared_resource (id, value) VALUES (1, 100);

-- Define the lock key
\set lock_key 12345

-- Acquire advisory lock and update value
SELECT pg_advisory_lock(:lock_key);

UPDATE shared_resource SET value = value + 10 WHERE id = 1;

-- Release advisory lock
-- Your code here

Hint

Use SELECT pg_advisory_unlock(:lock_key) to release the lock.

Practice

1. What is the main purpose of advisory locks in PostgreSQL?

easy

A. To control access to resources using user-defined keys

B. To automatically manage table-level locks during transactions

C. To speed up query execution by caching results

D. To backup the database safely without downtime

Solution

1. Step 1: Understand advisory locks concept

   Advisory locks allow applications to coordinate access to resources by using custom keys, not automatic locks on tables or rows.

2. Step 2: Compare options

   The other options describe other database features unrelated to advisory locks.

3. Final Answer:

   To control access to resources using user-defined keys -> Option A

4. Quick Check:

   Advisory locks = user-defined resource control [OK]

Hint: Advisory locks use keys to manage resource access [OK]

Common Mistakes:

• Confusing advisory locks with automatic table locks
• Thinking advisory locks speed up queries
• Assuming advisory locks handle backups

2. Which of the following is the correct syntax to acquire a session-level advisory lock with key 12345?

easy

A. SELECT pg_advisory_lock(12345);

B. LOCK TABLE pg_advisory_lock(12345);

C. SELECT acquire_lock(12345);

D. BEGIN LOCK 12345;

Solution

1. Step 1: Recall advisory lock syntax

   PostgreSQL uses the function pg_advisory_lock(key) to acquire a session-level advisory lock.

2. Step 2: Evaluate options

   SELECT pg_advisory_lock(12345); is the correct function call. The other options use invalid syntax or non-existent functions.

3. Final Answer:

   SELECT pg_advisory_lock(12345); -> Option A

4. Quick Check:

   pg_advisory_lock(key) = correct syntax [OK]

Hint: Use SELECT pg_advisory_lock(key) to lock [OK]

Common Mistakes:

• Using LOCK TABLE instead of function call
• Calling non-existent functions like acquire_lock
• Trying to lock with BEGIN LOCK syntax

3. What will be the result of this query if the advisory lock with key 999 is already held by another session?

SELECT pg_try_advisory_lock(999);

medium

A. true

B. false

C. null

D. error

Solution

1. Step 1: Understand pg_try_advisory_lock behavior

   This function tries to acquire the lock immediately and returns true if successful, false if the lock is held by someone else.

2. Step 2: Analyze the scenario

   Since the lock with key 999 is already held, the function returns false without waiting.

3. Final Answer:

   false -> Option B

4. Quick Check:

   pg_try_advisory_lock returns false if lock busy [OK]

Hint: pg_try_advisory_lock returns false if lock busy [OK]

Common Mistakes:

• Expecting true even if lock is held
• Thinking it returns null or error
• Confusing pg_try_advisory_lock with pg_advisory_lock

4. You wrote this code:

SELECT pg_advisory_unlock(123);

But the lock was never acquired before. What will happen?

medium

A. The function returns true and releases the lock

B. The function blocks until the lock is acquired

C. The function throws an error

D. The function returns false because no lock was held

Solution

1. Step 1: Understand pg_advisory_unlock behavior

   This function releases a lock if held and returns true; if no lock was held, it returns false.

2. Step 2: Analyze the scenario

   Since the lock was never acquired, the function returns false without error or blocking.

3. Final Answer:

   The function returns false because no lock was held -> Option D

4. Quick Check:

   Unlock returns false if lock not held [OK]

Hint: Unlock returns false if no lock held [OK]

Common Mistakes:

• Expecting an error when unlocking unheld lock
• Thinking unlock blocks or waits
• Assuming unlock always returns true

5. You want to ensure two different sessions do not run a critical section simultaneously using advisory locks. Which approach is best?

-- Session 1 and 2 run this code:
SELECT pg_try_advisory_lock(42);
-- If true, run critical section, then
SELECT pg_advisory_unlock(42);

hard

A. Use pg_advisory_unlock before acquiring lock to clear old locks

B. Use pg_try_advisory_lock to attempt lock and skip if busy

C. Use pg_advisory_lock to wait until lock is available before running

D. Use random keys each time to avoid conflicts

Solution

1. Step 1: Understand locking strategies

   pg_try_advisory_lock returns immediately and may skip critical section if lock busy; pg_advisory_lock waits until lock is free.

2. Step 2: Choose best approach for critical section

   To ensure only one session runs critical section at a time, waiting for the lock is safer than skipping it.

3. Step 3: Evaluate other options

   Unlocking before acquiring is unsafe and random keys defeat locking purpose.

4. Final Answer:

   Use pg_advisory_lock to wait until lock is available before running -> Option C

5. Quick Check:

   Waiting lock ensures exclusive access [OK]

Hint: Use pg_advisory_lock to wait for exclusive access [OK]

Common Mistakes:

• Using try lock and skipping critical section silently
• Unlocking before locking without owning lock
• Using random keys causing no real locking