- About TiDB
- Quick Start
- Software and Hardware Requirements
- Environment Configuration Checklist
- Plan Cluster Topology
- Install and Start
- Verify Cluster Status
- Test Cluster Performance
- Migrate from MySQL
- Migrate from CSV Files
- Migrate from SQL Files
- Replicate Incremental Data between TiDB Clusters in Real Time
- Backup and Restore
- Configure Time Zone
- Daily Checklist
- Maintain TiFlash
- Maintain TiDB Using TiUP
- Modify Configuration Online
- Online Unsafe Recovery
- Monitor and Alert
- TiDB Troubleshooting Map
- Identify Slow Queries
- Analyze Slow Queries
- SQL Diagnostics
- Identify Expensive Queries
- Statement Summary Tables
- Troubleshoot Hotspot Issues
- Troubleshoot Increased Read and Write Latency
- Save and Restore the On-Site Information of a Cluster
- Troubleshoot Cluster Setup
- Troubleshoot High Disk I/O Usage
- Troubleshoot Lock Conflicts
- Troubleshoot TiFlash
- Troubleshoot Write Conflicts in Optimistic Transactions
- Performance Tuning
- System Tuning
- Software Tuning
- SQL Tuning
- Understanding the Query Execution Plan
- SQL Optimization Process
- Logic Optimization
- Physical Optimization
- Prepare Execution Plan Cache
- Control Execution Plans
- Multiple Data Centers in One City Deployment
- Three Data Centers in Two Cities Deployment
- Two Data Centers in One City Deployment
- Read Historical Data
- Best Practices
- Use Placement Rules
- Use Load Base Split
- Use Store Limit
- TiDB Tools
- Use Cases
- TiDB Operator
- Backup & Restore (BR)
- TiDB Binlog
- TiDB Lightning
- TiDB Data Migration
- Cluster Architecture
- Key Monitoring Metrics
- SQL Language Structure and Syntax
- SQL Statements
ADMIN CANCEL DDL
ADMIN CHECKSUM TABLE
ADMIN CHECK [TABLE|INDEX]
ADMIN SHOW DDL [JOBS|QUERIES]
ADMIN SHOW TELEMETRY
ALTER PLACEMENT POLICY
CREATE [GLOBAL|SESSION] BINDING
CREATE PLACEMENT POLICY
CREATE TABLE LIKE
DROP [GLOBAL|SESSION] BINDING
DROP PLACEMENT POLICY
SET DEFAULT ROLE
SET [NAMES|CHARACTER SET]
SET [GLOBAL|SESSION] <variable>
SHOW ANALYZE STATUS
SHOW [GLOBAL|SESSION] BINDINGS
SHOW CHARACTER SET
SHOW [FULL] COLUMNS FROM
SHOW CREATE PLACEMENT POLICY
SHOW CREATE SEQUENCE
SHOW CREATE TABLE
SHOW CREATE USER
SHOW DRAINER STATUS
SHOW [FULL] FIELDS FROM
SHOW INDEX [FROM|IN]
SHOW INDEXES [FROM|IN]
SHOW KEYS [FROM|IN]
SHOW MASTER STATUS
SHOW PLACEMENT FOR
SHOW PLACEMENT LABELS
SHOW [FULL] PROCESSSLIST
SHOW PUMP STATUS
SHOW TABLE NEXT_ROW_ID
SHOW TABLE REGIONS
SHOW TABLE STATUS
SHOW [FULL] TABLES
SHOW [GLOBAL|SESSION] VARIABLES
- Data Types
- Functions and Operators
- Type Conversion in Expression Evaluation
- Control Flow Functions
- String Functions
- Numeric Functions and Operators
- Date and Time Functions
- Bit Functions and Operators
- Cast Functions and Operators
- Encryption and Compression Functions
- Information Functions
- JSON Functions
- Aggregate (GROUP BY) Functions
- Window Functions
- Miscellaneous Functions
- Precision Math
- Set Operations
- List of Expressions for Pushdown
- TiDB Specific Functions
- Clustered Indexes
- Generated Columns
- SQL Mode
- Table Attributes
- Garbage Collection (GC)
- Temporary Tables
- Character Set and Collation
- Placement Rules in SQL
- System Tables
- TiDB Dashboard
- Overview Page
- Cluster Info Page
- Key Visualizer Page
- Metrics Relation Graph
- SQL Statements Analysis
- Slow Queries Page
- Cluster Diagnostics
- Search Logs Page
- Instance Profiling
- Session Management and Configuration
- Command Line Flags
- Configuration File Parameters
- System Variables
- Storage Engines
- Errors Codes
- Table Filter
- Schedule Replicas by Topology Labels
- Release Notes
- All Releases
- Release Timeline
To make the usage of TiDB closer to traditional databases and reduce the cost of migration, starting from v3.0, TiDB supports the pessimistic transaction mode on top of the optimistic transaction model. This document describes the features of the TiDB pessimistic transaction mode.
Starting from v3.0.8, newly created TiDB clusters use the pessimistic transaction mode by default. However, this does not affect your existing cluster if you upgrade it from v3.0.7 or earlier to v3.0.8 or later. In other words, only newly created clusters default to using the pessimistic transaction mode.
You can set the transaction mode by configuring the
tidb_txn_mode system variable. The following command sets all explicit transactions (that is, non-autocommit transactions) executed by newly created sessions in the cluster to the pessimistic transaction mode:
SET GLOBAL tidb_txn_mode = 'pessimistic';
You can also explicitly enable the pessimistic transaction mode by executing the following SQL statements:
BEGIN /*T! PESSIMISTIC */;
BEGIN PESSIMISTIC; and
BEGIN OPTIMISTIC; statements take precedence over the
tidb_txn_mode system variable. Transactions started with these two statements ignore the system variable and support using both the pessimistic and optimistic transaction modes.
Pessimistic transactions in TiDB behave similarly to those in MySQL. See the minor differences in Difference with MySQL InnoDB.
For pessimistic transactions, TiDB introduces snapshot read and current read.
Snapshot read: it is an unlocked read that reads a version committed before the transaction starts. The read in the
SELECTstatement is a snapshot read.
Current read: it is a locked read that reads the latest committed version. The read in the
SELECT FOR UPDATEstatement is a current read.
The following examples provide a detailed description of snapshot read and current read.
Session 1 Session 2 Session 3 CREATE TABLE t (a INT); INSERT INTO T VALUES(1); BEGIN PESSIMISTIC; UPDATE t SET a = a + 1; BEGIN PESSIMISTIC; SELECT * FROM t; -- Use the snapshot read to read the version committed before the current transaction starts. The result returns a=1. BEGIN PESSIMISTIC; SELECT * FROM t FOR UPDATE; -- Use the current read. Wait for the lock. COMMIT; -- Release the lock. The SELECT FOR UPDATE operation of session 3 obtains the lock and TiDB uses the current read to read the latest committed version. The result returns a=2. SELECT * FROM t; -- Use the snapshot read to read the version committed before the current transaction starts. The result returns a=1.
When you execute
INSERTstatements, the latest committed data is read, data is modified, and a pessimistic lock is applied on the modified rows.
SELECT FOR UPDATEstatements, a pessimistic lock is applied on the latest version of the committed data, instead of on the modified rows.
Locks will be released when the transaction is committed or rolled back. Other transactions attempting to modify the data are blocked and have to wait for the lock to be released. Transactions attempting to read the data are not blocked, because TiDB uses multi-version concurrency control (MVCC).
If several transactions are trying to acquire each other's respective locks, a deadlock will occur. This is automatically detected, and one of the transactions will randomly be terminated with a MySQL-compatible error code
Transactions will wait up to
innodb_lock_wait_timeoutseconds (default: 50) to acquire new locks. When this timeout is reached, a MySQL-compatible error code
1205is returned. If multiple transactions are waiting for the same lock, the order of priority is approximately based on the
start tsof the transaction.
TiDB supports both the optimistic transaction mode and pessimistic transaction mode in the same cluster. You can specify either mode for transaction execution.
TiDB supports the
FOR UPDATE NOWAITsyntax and does not block and wait for locks to be released. Instead, a MySQL-compatible error code
Batch Point Getoperators do not read data, they still lock the given primary key or unique key, which blocks other transactions from locking or writing data to the same primary key or unique key.
TiDB supports the
FOR UPDATE OF TABLESsyntax. For a statement that joins multiple tables, TiDB only applies pessimistic locks on the rows that are associated with the tables in
When TiDB executes DML or
SELECT FOR UPDATEstatements that use range in the WHERE clause, concurrent DML statements within the range are not blocked.
CREATE TABLE t1 ( id INT NOT NULL PRIMARY KEY, pad1 VARCHAR(100) ); INSERT INTO t1 (id) VALUES (1),(5),(10);
BEGIN /*T! PESSIMISTIC */; SELECT * FROM t1 WHERE id BETWEEN 1 AND 10 FOR UPDATE;
BEGIN /*T! PESSIMISTIC */; INSERT INTO t1 (id) VALUES (6); -- blocks only in MySQL UPDATE t1 SET pad1='new value' WHERE id = 5; -- blocks waiting in both MySQL and TiDB
This behavior is because TiDB does not currently support gap locking.
TiDB does not support
SELECT LOCK IN SHARE MODE.
SELECT LOCK IN SHARE MODEis executed, it has the same effect as that without the lock, so the read or write operation of other transactions is not blocked.
DDL may result in failure of the pessimistic transaction commit.
When DDL is executed in MySQL, it might be blocked by the transaction that is being executed. However, in this scenario, the DDL operation is not blocked in TiDB, which leads to failure of the pessimistic transaction commit:
ERROR 1105 (HY000): Information schema is changed. [try again later]. TiDB executes the
TRUNCATE TABLEstatement during the transaction execution, which might result in the
table doesn't existerror.
START TRANSACTION WITH CONSISTENT SNAPSHOT, MySQL can still read the tables that are created later in other transactions, while TiDB cannot.
The autocommit transactions prefer the optimistic locking.
When using the pessimistic model, the autocommit transactions first try to commit the statement using the optimistic model that has less overhead. If a write conflict occurs, the pessimistic model is used for transaction retry. Therefore, if
tidb_retry_limitis set to
0, the autocommit transaction still reports the
Write Conflicterror when a write conflict occurs.
SELECT FOR UPDATEstatement does not wait for lock.
The data read by
EMBEDDED SELECTin the statement is not locked.
Open transactions in TiDB do not block garbage collection (GC). By default, this limits the maximum execution time of pessimistic transactions to 1 hour. You can modify this limit by editing
[performance]in the TiDB configuration file.
TiDB supports the following two isolation levels in the pessimistic transaction mode:
Repeatable Read by default, which is the same as MySQL.Note
In this isolation level, DML operations are performed based on the latest committed data. The behavior is the same as MySQL, but differs from the optimistic transaction mode in TiDB. See Difference between TiDB and MySQL Repeatable Read.
Adding a pessimistic lock requires writing data into TiKV. The response of successfully adding a lock can only be returned to TiDB after commit and apply through Raft. Therefore, compared with optimistic transactions, the pessimistic transaction mode inevitably has higher latency.
To reduce the overhead of locking, TiKV implements the pipelined locking process: when the data meets the requirements for locking, TiKV immediately notifies TiDB to execute subsequent requests and writes into the pessimistic lock asynchronously. This process reduces most latency and significantly improves the performance of pessimistic transactions. However, when network partition occurs in TiKV or a TiKV node is down, the asynchronous write into the pessimistic lock might fail and affect the following aspects:
Other transactions that modify the same data cannot be blocked. If the application logic relies on locking or lock waiting mechanisms, the correctness of the application logic is affected.
There is a low probability that the transaction commit fails, but it does not affect the correctness of the transactions.
If the application logic relies on the locking or lock waiting mechanisms, or if you want to guarantee as much as possible the success rate of transaction commits even in the case of TiKV cluster anomalies, you should disable the pipelined locking feature.
This feature is enabled by default. To disable it, modify the TiKV configuration:
[pessimistic-txn] pipelined = false
If the TiKV cluster is v4.0.9 or later, you can also dynamically disable this feature by modifying TiKV configuration online:
set config tikv pessimistic-txn.pipelined='false';