CHANGELOG.md

TiDB Changelog

All notable changes to this project will be documented in this file. See also Release Notes, TiKV Changelog and PD Changelog.

[2.1.0-rc.5] - 2018-11-12

SQL Optimizer

• Fix the issue that IndexReader reads the wrong handle in some cases #8132
• Fix the issue occurred while the IndexScan Prepared statement uses Plan Cache #8055
• Fix the issue that the result of the Union statement is unstable #8165

SQL Execution Engine

• Improve the performance of TiDB on inserting or updating wide tables #8024
• Support the unsigned int flag in the Truncate built-in function #8068
• Fix the error occurred while converting JSON data to the decimal type #8109
• Fix the error occurred when you Update the float type #8170

Statistics

• Fix the incorrect statistics issue during point queries in some cases #8035
• Fix the selectivity estimation of statistics for primary key in some cases #8149
• Fix the issue that the statistics of deleted tables are not cleared up for a long period of time #8182

Server

• Improve the readability of logs and make logs better
  • #8063
  • #8053
  • #8224
• Fix the error occurred when obtaining the table data of infoschema.profiling #8096
• Replace the unix socket with the pumps client to write binlogs #8098
• Add the threshold value for the tidb_slow_log_threshold environment variable, which dynamically sets the slow log #8094
• Add the original length of a SQL statement truncated while the tidb_query_log_max_len environment variable dynamically sets logs 8200
• Add the tidb_opt_write_row_id environment variable to control whether to allow writing _tidb_rowid #8218
• Add an upper bound to the Scan command of ticlient, to avoid overbound scan #8081, #8247

DDL

• Fix the issue that executing DDL statements in transactions encounters an error in some cases #8056
• Fix the issue that executing truncate table in partition tables does not take effect #8103
• Fix the issue that the DDL operation does not roll back correctly after being cancelled in some cases #8057
• Add the admin show next_row_id command to return the next available row ID #8268

[2.1.0-rc.4] - 2018-10-23

SQL Optimizer

• Fix the issue that column pruning of UnionAll is incorrect in some cases #7941
• Fix the issue that the result of the UnionAll operator is incorrect in some cases #8007

SQL Execution Engine

• Fix the precision issue of the AVG function #7874
• Support using the EXPLAIN ANALYZE statement to check the runtime statistics including the execution time and the number of returned rows of each operator during the query execution process #7925
• Fix the panic issue of the PointGet operator when a column of a table appears multiple times in the result set #7943
• Fix the panic issue caused by too large values in the Limit subclause #8002
• Fix the panic issue during the execution process of the AddDate/SubDate statement in some cases #8009

Statistics

• Fix the issue of judging the prefix of the histogram low-bound of the combined index as out of range #7856
• Fix the memory leak issue caused by statistics collecting #7873
• Fix the panic issue when the histogram is empty #7928
• Fix the issue that the histogram bound is out of range when the statistics is being uploaded #7944
• Limit the maximum length of values in the statistics sampling process #7982

Server

• Refactor Latch to avoid misjudgment of transaction conflicts and improve the execution performance of concurrent transactions #7711
• Fix the panic issue caused by collecting slow queries in some cases #7874
• Fix the panic issue when ESCAPED BY is an empty string in the LOAD DATA statement #8005
• Complete the “coprocessor error” log information #8006

Compatibility

• Set the Command field of the SHOW PROCESSLIST result to Sleep when the query is empty #7839

Expressions

• Fix the constant folding issue of the SYSDATE function #7895
• Fix the issue that SUBSTRING_INDEX panics in some cases #7897

DDL

• Fix the stack overflow issue caused by throwing the invalid ddl job type error #7958
• Fix the issue that the result of ADMIN CHECK TABLE is incorrect in some cases #7975

[2.1.0-rc.2] - 2018-09-14

SQL Optimizer

• Put forward a proposal of the next generation Planner #7543
• Improve the optimization rules of constant propagation #7276
• Enhance the computing logic of Range to enable it to handle multiple IN or EQUAL conditions simultaneously #7577
• Fix the issue that the estimation result of TableScan is incorrect when Range is empty #7583
• Support the PointGet operator for the UPDATE statement #7586
• Fix the panic issue during the process of executing the FirstRow aggregate function in some conditions #7624

SQL Execution Engine

• Fix the potential DataRace issue when the HashJoin operator encounters an error #7554
• Make the HashJoin operator read the inner table and build the hash table simultaneously #7544
• Optimize the performance of Hash aggregate operators #7541
• Optimize the performance of Join operators #7493, #7433
• Fix the issue that the result of UPDATE JOIN is incorrect when the Join order is changed #7571
• Improve the performance of Chunk’s iterator #7585

Statistics

• Fix the issue that the auto Analyze work repeatedly analyzes the statistics #7550
• Fix the statistics update error that occurs when there is no statistics change #7530
• Use the RC isolation level and low priority when building Analyze requests #7496
• Support enabling statistics auto-analyze on certain period of a day #7570
• Fix the panic issue when logging the statistics information #7588
• Support configuring the number of buckets in the histogram using the ANALYZE TABLE WITH BUCKETS statement #7619
• Fix the panic issue when updating an empty histogram #7640
• Update information_schema.tables.data_length using the statistics information #7657

Server

• Add Trace related dependencies #7532
• Enable the mutex profile feature of Golang #7512
• The Admin statement requires the Super_priv privilege #7486
• Forbid users to Drop crucial system tables #7471
• Switch from juju/errors to pkg/errors #7151
• Complete the functional prototype of SQL Tracing #7016
• Remove the goroutine pool #7564
• Support viewing the goroutine information using the USER1 signal #7587
• Set the internal SQL to high priority while TiDB is started #7616
• Use different labels to filter internal SQL and user SQL in monitoring metrics #7631
• Store the top 30 slow queries in the last week to the TiDB server #7646
• Put forward a proposal of setting the global system time zone for the TiDB cluster #7656
• Enrich the error message of “GC life time is shorter than transaction duration” #7658
• Set the global system time zone when starting the TiDB cluster #7638

Compatibility

• Add the unsigned flag for the Year type #7542
• Fix the issue of configuring the result length of the Year type in the Prepare/Execute mode #7525
• Fix the issue of inserting zero timestamp in the Prepare/Execute mode #7506
• Fix the error handling issue of the integer division #7492
• Fix the compatibility issue when processing ComStmtSendLongData #7485
• Fix the error handling issue during the process of converting string to integer #7483
• Optimize the accuracy of values in the information_schema.columns_in_table table #7463
• Fix the compatibility issue when writing or updating the string type of data using the MariaDB client #7573
• Fix the compatibility issue of aliases of the returned value #7600
• Fix the issue that the NUMERIC_SCALE value of the float type is incorrect in the information_schema.COLUMNS table #7602
• Fix the issue that Parser reports an error when the single line comment is empty #7612

Expressions

• Check the value of max_allowed_packet in the insert function #7528
• Support the built-in function json_contains #7443
• Support the built-in function json_contains_path #7596
• Support the built-in function encode/decode #7622
• Fix the issue that some time related functions are not compatible with the MySQL behaviors in some cases #7636
• Fix the compatibility issue of parsing the time type of data in string #7654
• Fix the issue that the time zone is not considered when computing the default value of the DateTime data #7655

DML

• Set correct last_insert_id in the InsertOnDuplicateUpdate statement #7534
• Reduce the cases of updating the auto_increment_id counter #7515
• Optimize the error message of Duplicate Key #7495
• Fix the insert...select...on duplicate key update issue #7406
• Support the LOAD DATA IGNORE LINES statement #7576

DDL

• Add the DDL job type and the current schema version information in the monitor #7472
• Complete the design of the Admin Restore Table feature #7383
• Fix the issue that the default value of the Bit type exceeds 128 #7249
• Fix the issue that the default value of the Bit type cannot be NULL #7604
• Reduce the interval of checking CREATE TABLE/DATABASE in the DDL queue #7608
• Use the ddl/owner/resign HTTP interface ro release the DDL owner and start electing a new owner #7649

TiKV Go Client

• Support the issue that the Seek operation only obtains Key #7419

Table Partition (Experimental)

• Fix the issue that the Bigint type cannot be used as the partition key #7520
• Support the rollback operation when an issue occurs during adding an index in the partitioned table #7437

[2.0.0-rc.5] - 2018-04-17

New Features

• Support showing memory usage of the executing statements in the SHOW PROCESSLIST statement
• Support setting the table comment using the Alter statement

Improvements

• Clean up the written data while rolling back the Add Index operation, to reduce consumed space
• Optimize the insert on duplicate key update statement to improve the performance by 10 times

Bug Fixes

• Fix the issue about applying the Top-N pushdown rule
• Fix the issue that Alter Table Modify Column reports an error in extreme conditions
• Fix the issue about the type of the results returned by the UNIX_TIMESTAMP function
• Fix the issue that the NULL value is inserted while adding NOT NULL columns
• Fix the estimation of the number of rows for the columns that contain NULL values
• Fix the zero value of the Binary type
• Fix the BatchGet issue within a transaction

[2.0.0-rc.4] - 2018-04-01

New Features

• Support SHOW GRANTS FOR CURRENT_USER();
• Support the SET TRANSACTION syntax
• Support displaying floating point numbers using scientific notation

Improvements

• Improve the execution performance of DecodeBytes
• Optimize LIMIT 0 to TableDual, to avoid building useless execution plans

Bug Fixes

• Fix the issue that the Expression in UnionScan is not cloned
• Fix the potential goroutine leak issue in copIterator
• Fix the issue that admin check table misjudges the unique index including null
• Fix the type inference issue during binary literal computing
• Fix the issue in parsing the CREATE VIEW statement
• Fix the panic issue when one statement contains both ORDER BY and LIMIT 0

[2.0.0-rc.3] - 2018-03-23

New Features

• Support closing the Join Reorder optimization in the optimizer using STRAIGHT_JOIN
• Output more detailed status information of DDL jobs in ADMIN SHOW DDL JOBS
• Support querying the original statements of currently running DDL jobs using ADMIN SHOW DDL JOB QUERIES
• Support recovering the index data using ADMIN RECOVER INDEX for disaster recovery
• Attach a lower priority to the ADD INDEX operation to reduce the impact on online business
• Support aggregation functions with JSON type parameters, such as SUM/AVG
• Support modifying the lower_case_table_names system variable in the configuration file, to support the OGG data synchronization tool
• Support using implicit RowID in CRUD operations

Improvements

• Improve compatibility with the Navicat management tool
• Use the Stream Aggregation operator when the GROUP BY clause is empty, to increase the speed
• Optimize the execution speed of ADD INDEX to greatly increase the speed in some scenarios
• Optimize checks on length and precision of the floating point type, to improve compatibility with MySQL
• Improve the parsing error log of time type and add more error information
• Improve memory control and add statistics about IndexLookupExecutor memory

Bug Fixes

• Fix the wrong result issue of MAX/MIN in some scenarios
• Fix the issue that the result of Sort Merge Join does not show in order of Join Key in some scenarios
• Fix the error of comparison between uint and int in boundary conditions

[2.0.0-rc.2] - 2018-03-15

Only TiKV has this release

[2.0.0-rc.1] - 2018-03-09

New Features

• Support limiting the memory usage by a single SQL statement, to reduce the risk of OOM
• Support pushing the Stream Aggregate operator down to TiKV
• Support validating the configuration file
• Support obtaining the information of TiDB configuration through HTTP API

Improvements

• Improve the compatibility with Navicat
• Improve the optimizer and extract common expressions with multiple OR conditions, to choose better query plan
• Improve the optimizer and convert subqueries to Join operators in more scenarios, to choose better query plan
• Compatible with more MySQL syntax in Parser
• Resolve Lock in the Batch mode to increase the garbage collection speed
• Optimize the Add Index operation and give lower priority to all write and read operations, to reduce the impact on online business

Bug Fixes

• Fix the length of Boolean field to improve compatibility

[1.1.0-beta] - 2018-02-24

New Features

• Add more monitoring metrics and refine the log
• Add the tidb_config session variable to output the current TiDB configuration
• Support displaying the table creating time in information_schema

Improvements

• Compatible with more MySQL syntax
• Optimize queries containing the MaxOneRow operator
• Configure the size of intermediate result sets generated by Join, to further reduce the memory used by Join
• Optimize the query performance of the SQL engine to improve the test result of the Sysbench Select/OLTP by 10%
• Improve the computing speed of subqueries in the optimizer using the new execution engine; compared with TiDB 1.0, TiDB 1.1 Beta has great improvement in tests like TPC-H and TPC-DS

Bug Fixes

• Fix the panic issue in the Union and Index Join operators
• Fix the wrong result issue of the Sort Merge Join operator in some scenarios
• Fix the issue that the Show Index statement shows indexes that are in the process of adding
• Fix the failure of the Drop Stats statement

[1.0.8] - 2018-02-11

New Features

• Add limitation (Configurable, the default value is 5000) to the DML statements number within a transaction

Improvements

• Improve the stability of the GC process by ignoring the regions with GC errors
• Run GC concurrently to accelerate the GC process
• Provide syntax support for the CREATE INDEX statement
• Optimize the performance of the InsertIntoIgnore statement

Bug Fixes

• Fix issues in the Outer Join result in some scenarios
• Fix the issue in the ShardRowID option
• Fix an issue in the Table/Column aliases returned by the Prepare statement
• Fix an issue in updating statistics delta
• Fix a panic error in the Drop Column statement
• Fix an DML issue when running the Add Column After statement

[1.0.7] - 2018-01-22

Improvements

• Optimize the FIELD_LIST command
• Fix data race of the information schema
• Avoid adding read-only statements to history
• Add the session variable to control the log query
• Add schema info API for the http status server
• Update the behavior when RunWorker is false in DDL
• Improve the stability of test results in statistics
• Support PACK_KEYS syntax for the CREATE TABLE statement
• Add row_id column for the null pushdown schema to optimize performance

Bug Fixes

• Fix the resource leak issue in statistics
• Fix the goroutine leak issue
• Fix an issue about IndexJoin

[1.1.0-alpha] - 2018-01-19

New Features

• Support the PROXY protocol

Improvements

• Support more syntax
• Reduce memory usage of statistics info using more compact structure
• Speed up loading statistics info when starting tidb-server
• Provide more accurate query cost evaluation
• Use Count-Min Sketch to estimate the cost of queries using unique index more accurately
• Support more complex conditions to make full use of index
• Refactor all executor operators using Chunk architecture, improve the execution performance of analytical statements and reduce memory usage
• Optimize performance of the INSERT IGNORE statement
• Push down more types and functions to TiKV
• Support more SQL_MODE
• Optimize the Load Data performance to increase the speed by 10 times
• Optimize the Use Database performance
• Support statistics on the memory usage of physical operators