Explicit locking

• YSQL

Row-level locks

YugabyteDB's YSQL supports explicit row-level locking, similar to PostgreSQL. Explicit row-locks ensure that two transactions can never hold conflicting locks on the same row. When two transactions try to acquire conflicting lock modes, the semantics are dictated by YugabyteDB's concurrency control policies.

The following types of row locks are supported:

• FOR UPDATE
• FOR NO KEY UPDATE
• FOR SHARE
• FOR KEY SHARE

The following example uses the FOR UPDATE row lock with the fail-on-conflict concurrency control policy. First, a row is selected for update, thereby locking it, and subsequently updated. A concurrent transaction should not be able to abort this transaction by updating the value of that row after the row is locked.

Create a sample table and populate it with sample data, as follows:

yugabyte=# CREATE TABLE t (k VARCHAR, v VARCHAR);
yugabyte=# INSERT INTO t VALUES ('k1', 'v1');

Next, connect to the universe using two independent ysqlsh instances. You can connect both session ysqlsh instances to the same server or to different servers.

Begin a transaction in the first session and perform a SELECT FOR UPDATE on the row in the table t. This locks the row for an update as a part of a transaction that has a very high priority (that is, in the high priority bucket, as explained in Transaction priorities):

yugabyte=# BEGIN TRANSACTION ISOLATION LEVEL REPEATABLE READ;

BEGIN

yugabyte=# SELECT * from t WHERE k='k1' FOR UPDATE;

 k  | v
----+----
 k1 | v1
(1 row)

Before completing the transaction, try to update the same key in your other session using a basic update statement, as follows:

yugabyte=# UPDATE t SET v='v1.1' WHERE k='k1';

ERROR:  All transparent retries exhausted. Operation failed. Try again: bb3aace4-5de2-41f9-981e-d9ca06671419 Conflicts with higher priority transaction: d4dadbf8-ca81-4bbd-b68c-067023f8ee6b

This operation fails because it conflicts with the row-level lock and as per Fail-on-Conflict concurrency control policy, the transaction aborts itself because it has a lower priority.

Note that the error message appears after all best-effort statement retries have been exhausted.

Finally, in the first session, update the row and commit the transaction, as follows:

yugabyte=# UPDATE t SET v='v1.2' WHERE k='k1';

UPDATE 1

yugabyte=# COMMIT;

COMMIT

This should succeed.

Advisory locks

TP Advisory locks are available in v2.25.1.0 and later.

YSQL also supports advisory locks, where the application manages concurrent access to resources through a cooperative locking mechanism. Advisory locks can be less resource-intensive than table or row locks for certain use cases because they don't involve scanning tables or indexes for lock conflicts. They are session-specific and managed by the client application.

In PostgreSQL, if an advisory lock is taken on one session, all sessions should be able to see the advisory locks acquired by any other session. Similarly, in YugabyteDB, if an advisory lock is acquired on one session, all the sessions should be able to see the advisory locks regardless of the node the session is connected to. This is achieved via the pg_advisory_locks system table, which is dedicated to hosting advisory locks. All advisory lock requests are stored in this system table.

To enable the use of advisory locks in a cluster, you must set the Advisory lock flags.

Using advisory locks

Advisory locks in YugabyteDB are semantically identical to PostgreSQL, and are managed using the same functions. Refer to Advisory lock functions in the PostgreSQL documentation.

You can acquire an advisory lock in the following ways:

• Session level

  Once acquired at session level, the advisory lock is held until it is explicitly released or the session ends.

  Unlike standard lock requests, session-level advisory lock requests do not honor transaction semantics: a lock acquired during a transaction that is later rolled back will still be held following the rollback, and likewise an unlock is effective even if the calling transaction fails later. A lock can be acquired multiple times by its owning process; for each completed lock request there must be a corresponding unlock request before the lock is actually released.

  SELECT pg_advisory_lock(10);
  

• Transaction level

  Transaction-level lock requests, on the other hand, behave more like regular row-level lock requests: they are automatically released at the end of the transaction, and there is no explicit unlock operation. This behavior is often more convenient than the session-level behavior for short-term usage of an advisory lock.

  SELECT pg_advisory_xact_lock(10);
  

Advisory locks can also be exclusive or shared:

• Exclusive Lock

  Only one session/transaction can hold the lock at a time. Other sessions/transactions can't acquire the lock until the lock is released.

  select pg_advisory_lock(10);
  select pg_advisory_xact_lock(10);
  

• Shared Lock

  Multiple sessions/transactions can hold the lock simultaneously. However, no session/transaction can acquire an exclusive lock while shared locks are held.

  select pg_advisory_lock_shared(10); 
  select pg_advisory_xact_lock_shared(10);
  

Finally, advisory locks can be blocking or non-blocking:

• Blocking lock

  The process trying to acquire the lock waits until the lock is acquired.

  select pg_advisory_lock(10); 
  select pg_advisory_xact_lock(10);
  

• Non-blocking lock

  The process immediately returns a boolean value stating if the lock is acquired or not.

  select pg_try_advisory_lock(10);
  select pg_try_advisory_xact_lock(10);