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.

SQL statements

Each of the following SQL statements control transactions in some way.

Syntax

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

To use client-side transaction retries, you should also include the SAVEPOINT cockroach_restart, ROLLBACK TO SAVEPOINT cockroach_restart 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 contention

Transactions in CockroachDB lock data resources that are written during their execution. When a pending write from one transaction conflicts with a write of a concurrent transaction, the concurrent transaction must wait for the earlier transaction to complete before proceeding. When a dependency cycle is detected between transactions, the transaction with the higher priority aborts the dependent transaction to avoid deadlock, which must be retried.

For more details about transaction contention and best practices for avoiding contention, see Understanding and Avoiding Transaction Contention.

Transaction retries

Transactions may require retries if they experience deadlock or read/write contention with other concurrent transactions which cannot be resolved without allowing potential serializable anomalies. (However, it's possible to mitigate read-write conflicts by performing reads using AS OF SYSTEM TIME.)

There are two cases for handling transaction retries:

• Automatic retries, which CockroachDB processes for you.
• Client-side intervention, which your application must handle.

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. 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.

In future versions of CockroachDB, we plan on providing stronger guarantees for read-only queries that return at most one row, regardless of the size of that row.

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 if the accumulated results overflow an internal buffer, 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. The the client/driver is using the PostgreSQL Simple Query protocol, a batch is made up 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;"
   )
   

Client-side intervention

Your application should include client-side retry handling when the statements are sent individually, such as:

> BEGIN;

> UPDATE products SET inventory = 0 WHERE sku = '8675309';

> INSERT INTO orders (customer, status) VALUES (1, 'new');

> COMMIT;

To indicate a transaction must be retried, CockroachDB surfaces an error with the code 40001 and an error message that begins with the string retry transaction.

To handle these types of errors you have two options:

• Recommended: Use the SAVEPOINT cockroach_restart functions to create retryable transactions. Retryable transactions can improve performance because their priority's increased each time they are retried, making them more likely to succeed the longer they're in your system.

  For more information, see Client-Side Transaction Retries.

• Abort the transaction using ROLLBACK, and then reissue all of the statements in the transaction. This does not automatically increase the transaction's priority, so it's possible in high-contention workloads for transactions to take an incredibly long time to succeed.

Client-side transaction retries

To improve the performance of transactions that fail due to contention, CockroachDB includes a set of statements that let you retry those transactions. Retrying transactions has the benefit of increasing their priority each time they're retried, increasing their likelihood to succeed.

Retried transactions are also issued at a later timestamp, so the transaction now operates on a later snapshot of the database, so the reads might return updated data.

Implementing client-side retries requires three statements:

• SAVEPOINT cockroach_restart declares the client's intent to retry the transaction if there are contention errors. It must be executed after BEGIN but before the first statement that manipulates a database.

• ROLLBACK TO SAVEPOINT cockroach_restart is used when your application detects 40001 / retry transaction errors. It provides you a chance to "retry" the transaction by rolling the database's state back to the beginning of the transaction and increasing the transaction's priority.

  After issuing ROLLBACK TO SAVEPOINT cockroach_restart, you must issue any statements you want the transaction to contain. Typically, this means recalculating values and reissuing a similar set of statements to the previous attempt.

• RELEASE SAVEPOINT cockroach_restart commits the transaction. At this point, CockroachDB checks to see if the transaction contends with others for access to the same values; the highest priority transaction succeeds, and the others return 40001 / retry transaction errors.

  You must also execute COMMIT afterward to clear the connection for the next transaction.

You can find examples of this in the Syntax section of this page or in our Build an App with CockroachDB tutorials.

It's also important to note that retried transactions are restarted at a later timestamp. This means that the transaction operates on a later snapshot of the database and related reads might retrieve updated data.

For greater detail, here's the process a retryable transaction goes through.

1. The transaction starts with the BEGIN statement.

2. The SAVEPOINT cockroach_restart statement declares the intention to retry the transaction in the case of contention errors. Note that CockroachDB's savepoint implementation does not support all savepoint functionality, such as nested transactions.

3. The statements in the transaction are executed.

4. If a statement returns a retryable error (identified via the 40001 error code or retry transaction string at the start of the error message), you can issue the ROLLBACK TO SAVEPOINT cockroach_restart statement to restart the transaction. Alternately, the original SAVEPOINT cockroach_restart statement can be reissued to restart the transaction.

   You must now issue the statements in the transaction again.

   In cases where you do not want the application to retry the transaction, you can simply issue ROLLBACK at this point. Any other statements will be rejected by the server, as is generally the case after an error has been encountered and the transaction has not been closed.

5. Once the transaction executes all statements without encountering contention errors, execute RELEASE SAVEPOINT cockroach_restart to commit the changes. If this succeeds, all changes made by the transaction become visible to subsequent transactions and are guaranteed to be durable if a crash occurs.

   In some cases, the RELEASE SAVEPOINT statement itself can fail with a retryable error, mainly because transactions in CockroachDB only realize that they need to be restarted when they attempt to commit. If this happens, the retryable error is handled as described in step 4.

Transaction priorities

Every transaction in CockroachDB is assigned an initial priority. By default, that priority is NORMAL, but for transactions that should be given preference in high-contention scenarios, the client can set the priority within the BEGIN statement:

> BEGIN PRIORITY <LOW | NORMAL | HIGH>;

Alternately, the client can set the priority immediately after the transaction is started as follows:

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

The client can also display the current priority of the transaction with SHOW TRANSACTION PRIORITY.

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, recent research has demonstrated that the use of weak isolation levels results in substantial vulnerability to concurrency-based attacks.

New in v2.1: CockroachDB now only supports SERIALIZABLE isolation. In previous versions of CockroachDB, you could set transactions to SNAPSHOT isolation, but that feature has been removed.

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.

See also

• BEGIN
• COMMIT
• ROLLBACK
• SAVEPOINT
• RELEASE SAVEPOINT
• SHOW
• Retryable function code samples

Was this helpful?