Row-level locking (FOR UPDATE, FOR SHARE) in PostgreSQL

Introduction

Row-level locking helps prevent conflicts when multiple people try to change the same data at the same time. It locks specific rows so others can't change them until you're done.

When you want to update a row but ensure no one else changes it until your update finishes.

When you want to read data and make sure it doesn't change while you are working with it.

When you are handling money transfers and need to lock account rows to avoid errors.

When multiple users might edit the same record and you want to avoid overwriting each other's changes.

Syntax

PostgreSQL

SELECT columns FROM table WHERE condition FOR UPDATE;

SELECT columns FROM table WHERE condition FOR SHARE;

FOR UPDATE locks the selected rows for writing (updates, deletes).

FOR SHARE locks the selected rows for reading, allowing others to read but not update them.

Examples

This locks the account with id 1 so no other transaction can update or delete it until this transaction finishes.

PostgreSQL

SELECT * FROM accounts WHERE id = 1 FOR UPDATE;

This locks the rows with pending orders for reading, so others can also read but cannot update or delete them until the lock is released.

PostgreSQL

SELECT * FROM orders WHERE status = 'pending' FOR SHARE;

Starts a transaction, locks product with id 10 for update, then commits changes releasing the lock.

PostgreSQL

BEGIN;
SELECT * FROM products WHERE id = 10 FOR UPDATE;
-- do some updates
COMMIT;

Sample Program

This example locks the employee row with id 5 so no other transaction can change it while we update the salary. After the update, the lock is released with COMMIT.

PostgreSQL

BEGIN;
SELECT * FROM employees WHERE id = 5 FOR UPDATE;
-- Imagine we update the employee's salary here
UPDATE employees SET salary = salary + 1000 WHERE id = 5;
COMMIT;

Important Notes

Row-level locks only last during the transaction. Always use BEGIN and COMMIT to control locking.

FOR UPDATE blocks other transactions trying to update or delete the locked rows.

FOR SHARE allows others to read but prevents updates or deletes until the lock is released.

Summary

Row-level locking controls access to specific rows during transactions.

Use FOR UPDATE to lock rows for writing.

Use FOR SHARE to lock rows for reading.

Practice

1. What does the FOR UPDATE clause do in a PostgreSQL query?

easy

A. Locks selected rows to prevent other transactions from modifying them until the current transaction ends.

B. Locks the entire table to prevent any access by other transactions.

C. Allows other transactions to read but not modify the selected rows.

D. Unlocks rows that were previously locked by another transaction.

Solution

1. Step 1: Understand the purpose of FOR UPDATE

   The FOR UPDATE clause locks the rows returned by the query to prevent other transactions from modifying them.

2. Step 2: Compare with other locking types

   Unlike FOR SHARE, which allows reading but prevents writing, FOR UPDATE blocks other transactions from modifying the locked rows until the current transaction ends.

3. Final Answer:

   Locks selected rows to prevent other transactions from modifying them until the current transaction ends. -> Option A

4. Quick Check:

   FOR UPDATE = exclusive row lock [OK]

Hint: FOR UPDATE locks rows for writing, blocking others [OK]

Common Mistakes:

• Confusing FOR UPDATE with table-level locks
• Thinking FOR UPDATE allows other writes
• Mixing FOR UPDATE with FOR SHARE behavior

2. Which of the following is the correct syntax to lock rows for reading using row-level locking in PostgreSQL?

easy

A. SELECT * FROM orders FOR UPDATE;

B. SELECT * FROM orders FOR SHARE;

C. SELECT * FROM orders LOCK ROW;

D. SELECT * FROM orders WITH LOCK;

Solution

1. Step 1: Identify the clause for read locks

   In PostgreSQL, FOR SHARE is used to lock rows for reading, allowing other transactions to read but not modify.

2. Step 2: Check syntax correctness

   Options B and C are invalid syntax. SELECT * FROM orders FOR UPDATE; locks rows for writing, not reading.

3. Final Answer:

   SELECT * FROM orders FOR SHARE; -> Option B

4. Quick Check:

   Read lock = FOR SHARE [OK]

Hint: FOR SHARE locks rows for reading, syntax: SELECT ... FOR SHARE [OK]

Common Mistakes:

• Using FOR UPDATE instead of FOR SHARE for read locks
• Using non-existent LOCK ROW or WITH LOCK syntax
• Confusing table-level locks with row-level locks

3. Consider the following two transactions running concurrently:

-- Transaction 1
BEGIN;
SELECT * FROM products WHERE id = 10 FOR UPDATE;
-- Transaction 2
BEGIN;
SELECT * FROM products WHERE id = 10 FOR SHARE;

What will happen when Transaction 2 tries to execute its SELECT statement?

medium

A. Transaction 2 will immediately acquire the FOR SHARE lock and proceed.

B. Transaction 2 will cause a deadlock error immediately.

C. Transaction 2 will wait until Transaction 1 commits or rolls back before proceeding.

D. Transaction 2 will ignore the lock and read the row without waiting.

Solution

1. Step 1: Understand locking conflict between FOR UPDATE and FOR SHARE

   A FOR UPDATE lock is exclusive and blocks other transactions from acquiring conflicting locks, including FOR SHARE.

2. Step 2: Analyze Transaction 2 behavior

   Transaction 2's FOR SHARE lock request conflicts with Transaction 1's FOR UPDATE lock, so it must wait until Transaction 1 finishes.

3. Final Answer:

   Transaction 2 will wait until Transaction 1 commits or rolls back before proceeding. -> Option C

4. Quick Check:

   FOR UPDATE blocks FOR SHARE until commit [OK]

Hint: FOR UPDATE blocks FOR SHARE until transaction ends [OK]

Common Mistakes:

• Assuming FOR SHARE can proceed during FOR UPDATE lock
• Expecting deadlock without waiting
• Ignoring lock conflicts between FOR UPDATE and FOR SHARE

4. You wrote this query to lock rows for update:

SELECT * FROM customers FOR SHARE;

But you want to prevent other transactions from modifying these rows. What is wrong and how to fix it?

medium

A. FOR SHARE locks the entire table; use WHERE clause to limit rows.

B. FOR SHARE is deprecated; use LOCK TABLE instead.

C. FOR SHARE requires an explicit NOWAIT clause to lock rows.

D. FOR SHARE locks rows only for reading; replace it with FOR UPDATE to lock for writing.

Solution

1. Step 1: Identify the locking behavior of FOR SHARE

   FOR SHARE locks rows to allow reading but does not prevent other transactions from modifying them.

2. Step 2: Choose correct clause to prevent modifications

   To block other transactions from modifying rows, use FOR UPDATE instead of FOR SHARE.

3. Final Answer:

   FOR SHARE locks rows only for reading; replace it with FOR UPDATE to lock for writing. -> Option D

4. Quick Check:

   Prevent writes = use FOR UPDATE [OK]

Hint: Use FOR UPDATE to block writes, not FOR SHARE [OK]

Common Mistakes:

• Thinking FOR SHARE blocks writes
• Using LOCK TABLE unnecessarily
• Missing WHERE clause but unrelated to locking type

5. You want to update multiple rows in a table but ensure no other transaction can read or modify these rows until your transaction finishes. Which locking clause should you use in your SELECT statement before updating?

hard

A. SELECT * FROM table_name FOR UPDATE;

B. SELECT * FROM table_name FOR SHARE;

C. SELECT * FROM table_name FOR NO KEY UPDATE;

D. SELECT * FROM table_name FOR KEY SHARE;

Solution

1. Step 1: Understand the locking levels

   FOR UPDATE locks rows exclusively, blocking writes by others until commit, but does not block reads.

2. Step 2: Compare with other lock types

   FOR SHARE and FOR KEY SHARE allow reads; FOR NO KEY UPDATE is less restrictive and allows some concurrent updates.

3. Step 3: Choose the strictest lock to block writes

   Only FOR UPDATE fully blocks other transactions from modifying the rows, but it does not block reads.

4. Final Answer:

   SELECT * FROM table_name FOR UPDATE; -> Option A

5. Quick Check:

   Block writes = FOR UPDATE [OK]

Hint: FOR UPDATE blocks writes until commit [OK]

Common Mistakes:

• Choosing FOR SHARE or KEY SHARE which allow writes
• Using NO KEY UPDATE which is less strict
• Not understanding lock levels and their effects