Transactions

On this page

CockroachDB supports bundling multiple SQL statements into a single all-or-nothing transaction. Each transaction guarantees ACID semantics spanning arbitrary tables and rows, even when data is distributed. If a transaction succeeds, all mutations are applied together with virtual simultaneity. If any part of a transaction fails, the entire transaction is aborted, and the database is left unchanged. CockroachDB guarantees that while a transaction is pending, it is isolated from other concurrent transactions with serializable isolation.

For a detailed discussion of CockroachDB transaction semantics, see How CockroachDB Does Distributed Atomic Transactions and Serializable, Lockless, Distributed: Isolation in CockroachDB. The explanation of the transaction model described in this blog post is slightly out of date. See the Transaction Retries section for more details.

SQL statements

The following SQL statements control transactions.

Syntax

In CockroachDB, a transaction is set up by surrounding SQL statements with the BEGIN and COMMIT statements.

To use advanced client-side transaction retries, you should also include the SAVEPOINT, ROLLBACK TO SAVEPOINT and RELEASE SAVEPOINT statements.

> BEGIN;

> SAVEPOINT cockroach_restart;

<transaction statements>

> RELEASE SAVEPOINT cockroach_restart;

> COMMIT;

At any time before it's committed, you can abort the transaction by executing the ROLLBACK statement.

Clients using transactions must also include logic to handle retries.

Error handling

To handle errors in transactions, you should check for the following types of server-side errors:

Transaction retries

Transactions may require retries due to contention with another concurrent or recent transaction attempting to write to the same data.

There are two cases in which transaction retries can occur:

• Automatic retries, which CockroachDB silently processes for you.
• Client-side retries, which your application must handle after receiving a transaction retry error.

To reduce the need for transaction retries, see Reduce transaction contention.

Automatic retries

CockroachDB automatically retries individual statements (implicit transactions) and transactions sent from the client as a single batch, as long as the size of the results being produced for the client, including protocol overhead, is less than 16KiB by default. Once that buffer overflows, CockroachDB starts streaming results back to the client, at which point automatic retries cannot be performed any more. As long as the results of a single statement or batch of statements are known to stay clear of this limit, the client does not need to worry about transaction retries.

You can increase the occurrence of automatic retries as a way to minimize transaction retry errors:

• Send statements in transactions as a single batch. Batching allows CockroachDB to automatically retry a transaction when previous reads are invalidated at a pushed timestamp. When a multi-statement transaction is not batched, and takes more than a single round trip, CockroachDB cannot automatically retry the transaction. For an example showing how to break up large transactions in an application, see Break up large transactions into smaller units of work.

• Limit the size of the result sets of your transactions to under 16KB, so that CockroachDB is more likely to automatically retry when previous reads are invalidated at a pushed timestamp. When a transaction returns a result set over 16KB, even if that transaction has been sent as a single batch, CockroachDB cannot automatically retry the transaction. You can change the results buffer size for all new sessions using the sql.defaults.results_buffer.size cluster setting, or for a specific session using the results_buffer_size connection parameter.

Individual statements

Individual statements are treated as implicit transactions, and so they fall under the rules described above. If the results are small enough, they will be automatically retried. In particular, INSERT/UPDATE/DELETE statements without a RETURNING clause are guaranteed to have minuscule result sizes. For example, the following statement would be automatically retried by CockroachDB:

> DELETE FROM customers WHERE id = 1;

Batched statements

Transactions can be sent from the client as a single batch. Batching implies that CockroachDB receives multiple statements without being asked to return results in between them; instead, CockroachDB returns results after executing all of the statements, except when the accumulated results overflow the buffer mentioned above, in which case they are returned sooner and automatic retries can no longer be performed.

Batching is generally controlled by your driver or client's behavior. Technically, it can be achieved in two ways, both supporting automatic retries:

1. When the client/driver is using the PostgreSQL Extended Query protocol, a batch is made up of all queries sent in between two Sync messages. Many drivers support such batches through explicit batching constructs.

2. When the client/driver is using the PostgreSQL Simple Query protocol, a batch is made up of semicolon-separated strings sent as a unit to CockroachDB. For example, in Go, this code would send a single batch (which would be automatically retried):

   db.Exec(
     "BEGIN;
   
     DELETE FROM customers WHERE id = 1;
   
     DELETE orders WHERE customer = 1;
   
     COMMIT;"
   )
   

Within a batch of statements, CockroachDB infers that the statements are not conditional on the results of previous statements, so it can retry all of them. Of course, if the transaction relies on conditional logic (e.g., statement 2 is executed only for some results of statement 1), then the transaction cannot be all sent to CockroachDB as a single batch. In these common cases, CockroachDB cannot retry, say, statement 2 in isolation. Since results for statement 1 have already been delivered to the client by the time statement 2 is forcing the transaction to retry, the client needs to be involved in retrying the whole transaction and so you should write your transactions to use client-side retry handling.

The enable_implicit_transaction_for_batch_statements session variable defaults to true. This means that any batch of statements is treated as an implicit transaction, so the BEGIN/COMMIT commands are not needed to group all the statements in one transaction.

Bounded staleness reads

In the event bounded staleness reads are used along with either the with_min_timestamp function or the with_max_staleness function and the nearest_only parameter is set to true, the query will throw an error if it can't be served by a nearby replica.

Nested transactions

CockroachDB supports the nesting of transactions using savepoints. These nested transactions are also known as sub-transactions. Nested transactions can be rolled back without discarding the state of the entire surrounding transaction.

This can be useful in applications that abstract database access using an application development framework or ORM. Different components of the application can operate on different sub-transactions without having to know about each others' internal operations, while trusting that the database will maintain isolation between sub-transactions and preserve data integrity.

Just as COMMIT and ROLLBACK are used to commit and discard entire transactions, respectively, RELEASE SAVEPOINT and ROLLBACK TO SAVEPOINT are used to commit and discard nested transactions. This relationship is shown in the following table:

For more information, including examples showing how to use savepoints to create nested transactions, see the savepoints documentation.

Transaction priorities

Every transaction in CockroachDB is assigned an initial priority. By default, the transaction priority is NORMAL.

Set transaction priority

For transactions that should be given higher or lower preference in high-contention scenarios, you can set the priority in the BEGIN statement:

> BEGIN PRIORITY <LOW | NORMAL | HIGH>;

You can also set the priority immediately after a transaction is started:

> SET TRANSACTION PRIORITY <LOW | NORMAL | HIGH>;

To set the default transaction priority for all transactions in a session, use the default_transaction_priority session variable. For example:

> SET default_transaction_priority = 'high';

View transaction priority

transaction_priority is a read-only session variable.

To view the current priority of a transaction, use SHOW transaction_priority or SHOW TRANSACTION PRIORITY:

> SHOW transaction_priority;

  transaction_priority
------------------------
  high

> SHOW TRANSACTION PRIORITY;

  transaction_priority
------------------------
  high

Isolation levels

CockroachDB executes all transactions at the strongest ANSI transaction isolation level: SERIALIZABLE. All other ANSI transaction isolation levels (e.g., SNAPSHOT, READ UNCOMMITTED, READ COMMITTED, and REPEATABLE READ) are automatically upgraded to SERIALIZABLE. Weaker isolation levels have historically been used to maximize transaction throughput. However, ACIDRain: Concurrency-Related Attacks on Database-Backed Web Applications has demonstrated that the use of weak isolation levels results in substantial vulnerability to concurrency-based attacks.

For a detailed discussion of isolation in CockroachDB transactions, see Serializable, Lockless, Distributed: Isolation in CockroachDB.

Serializable isolation

With SERIALIZABLE isolation, a transaction behaves as though it has the entire database all to itself for the duration of its execution. This means that no concurrent writers can affect the transaction unless they commit before it starts, and no concurrent readers can be affected by the transaction until it has successfully committed. This is the strongest level of isolation provided by CockroachDB and it's the default.

SERIALIZABLE isolation permits no anomalies. To prevent write skew anomalies, SERIALIZABLE isolation may require transaction restarts. For a demonstration of SERIALIZABLE preventing write skew, see Serializable Transactions.

Comparison to ANSI SQL isolation levels

CockroachDB uses slightly different isolation levels than ANSI SQL isolation levels.

Aliases

SNAPSHOT, READ UNCOMMITTED, READ COMMITTED, and REPEATABLE READ are aliases for SERIALIZABLE.

Comparison

The CockroachDB SERIALIZABLE level is stronger than the ANSI SQL READ UNCOMMITTED, READ COMMITTED, and REPEATABLE READ levels and equivalent to the ANSI SQL SERIALIZABLE level.

For more information about the relationship between these levels, see A Critique of ANSI SQL Isolation Levels.

Limit the number of rows written or read in a transaction

You can limit the number of rows written or read in a transaction at the cluster or session level. This allows you configure CockroachDB to log or reject statements that could destabilize a cluster or violate application best practices.

Use the cluster settings sql.defaults.transaction_rows_written_log, sql.defaults.transaction_rows_written_err, sql.defaults.transaction_rows_read_log, and sql.defaults.transaction_rows_read_err and session settings transaction_rows_written_log, transaction_rows_written_err, transaction_rows_read_log, and transaction_rows_read_err to limit the number of rows written or read in a transaction. When the log limit is reached, the transaction is logged to the SQL_PERF channel. When the err limit is reached, the transaction is rejected. The limits are enforced after each statement of a transaction has been fully executed.

The "write" limits apply to INSERT, INSERT INTO SELECT FROM, INSERT ON CONFLICT, UPSERT, UPDATE, and DELETE SQL statements. The "read" limits apply to the SELECT statement in addition to the statements subject to the "write" limits. The limits do not apply to CREATE TABLE AS, SELECT, IMPORT, TRUNCATE, DROP, ALTER TABLE, BACKUP, RESTORE, or CREATE STATISTICS statements.

See also

• BEGIN
• COMMIT
• ROLLBACK
• SAVEPOINT
• RELEASE SAVEPOINT
• SHOW
• Retryable transaction example code in Java using JDBC
• DB Console Transactions Page
• CockroachDB Architecture: Transaction Layer
• Transaction retry error reference

Was this helpful?