Version guide for V5.3.x instances

Updated at:

This topic provides the version guide for PolarDB-X 1.0 V5.3.x instances.

v5.3.12-15708775 (October 12, 2019)

CategoryDescription
New featuresAdds a chunk-based executor to optimize complex queries.
Adds the SQL execution plan management feature.
Adds a new DDL task engine that supports viewing, recovering, and rolling back DDL tasks.
Adds support for specifying an index for a query using the FORCE INDEX hint.
Adds support for the ROW function in non-scalar subqueries.
Adds full support for escape characters in time functions such as DATE_FORMAT and INTERVAL.
Performance optimizationsEnables cost-based optimization (CBO) by default on primary instances to automatically select JOIN orders and operator implementations.
Optimizes more types of subqueries, such as SemiJoin, and their pushdown.
Merges transaction instructions with application business SQL execution to reduce backend database network communication and improve transaction performance.
Optimizes the parameter settings of the internal thread pool.
Optimizes the fast recovery of the Druid connection pool when a backend RDS instance is abnormal.
Fixed issuesFixes an issue with arithmetic precedence in SQL parsing.
Fixes a connection leak issue in some HASH JOIN scenarios.
Fixes an error that occurs in the physical SQL statement of a unary operator.
Fixes an issue where the LOCK TABLE statement cannot be run correctly on a single database.
Fixes an issue with SAVEPOINT support in PolarDB-X 1.0.
Fixes an incompatibility issue between the MD5 and HEX functions and MySQL.
Fixes an error that occurs when you run the SELECT USER(), DATABASE() LIMIT 1 statement.
Fixes an issue where the concurrent behavior of MERGE_UNION=FALSE does not meet expectations.
Fixes an issue where the query routing result is incorrect for a date range query with multiple AND operators.
Fixes an issue where the destination database name is ignored in a RENAME TABLE statement.
Fixes an issue where administrator accounts cannot view the permissions of all users.
Fixes a low-probability issue where the degree of parallelism cost is calculated inaccurately for a primary instance.
Fixes an abnormal log rotation issue caused by a cleanup script.
Fixes an error that occurs in the physical SQL statement pushed down when an UPDATE or DELETE statement for a single table contains a table name reference.
Fixes an error caused by memory limits in a LIMIT statement that does not contain an ORDER BY clause.
Fixes an issue where query results are inconsistent for a query with multiple AND conditions.
Fixes the error message for an UPDATE statement that contains a subquery.
Fixes a parsing issue that occurs when a function appears in both the DISTINCT and ORDER BY clauses.

v5.3.11-15663020 (August 20, 2019)

CategoryDescription
Fixed issuesFixes an issue where the monitoring of physical response time (RT) is inaccurate.
Adjusts the configuration of worker threads.

V5.3.11-15622313 (July 4, 2019)

CategoryDescription
Performance optimizationsOptimizes JOIN reordering and physical execution policies based on CBO for analytics-oriented read-only instances.
Optimizes the full table scan speed for analytics-oriented read-only instances.
Optimizes the performance of INSERT, REPLACE, UPDATE, and DELETE statements.
Fixed issuesFixes an issue where concurrent KILL commands prevent users from connecting to PolarDB-X 1.0.
Fixes an error that occurs during synchronization to PolarDB-X 1.0 when the auto-increment primary key of the source table has a value of 0.
Fixes an error that occurs in a multi-table UPDATE statement.
Fixes an error that occurs in a SHOW TABLES statement when a user has a view.
Fixes an issue where a table with a long name cannot be displayed or deleted in the console.
Fixes an issue where accounts with permissions on multiple databases cannot use Navicat concurrently.
Fixes an issue where a user query hangs if the result set is too large when the COMPRESS protocol is used.
Fixes a query error that occurs when the parameter of the AVG function is of the TIMESTAMP type.
Fixes an issue with the return value type of the ROUND function.
Fixes an issue where the number of reads and writes in SHOW NODE exceeds the integer limit.
Fixes some DDL and statement parsing errors.
Fixes an error that occurs when you create a broadcast table with a database name.
Fixes a sharding calculation exception caused by physical SQL optimization in some scenarios.
Fixes an issue where the SHOW GRANTS statement for an ADMIN account cannot display the permissions of all users.
Fixes an error that occurs when only one column has a value in a multi-column IN clause.
Fixes a system exception caused by a LOCK TABLES statement for a single table.
Fixes an error that occurs in an EXPLAIN statement when the shard key in an INSERT statement contains a function.

V5.3.10-15585948 (May 23, 2019)

CategoryDescription
New featuresAdds support for parallel queries.
Adds support for the UNSIGNED type in scalar correlated subqueries.
Adds cross-schema support for ALTER SEQUENCE.
Supports WHERE clauses in queries that use NEXTVAL for a sequence, such as SELECT test.NEXTVAL FROM DUAL WHERE COUNT=10.
Performance optimizationsOptimizes the performance of queries with shard keys.
Optimizes the connection pool feature by adding support for keep-alive.
Fixed issuesFixes an error that occurs when you run a DESC statement that contains a schema name, such as DESC DB.ORDER.
Fixes an error that occurs when you run the SHOW TABLE STATUS statement on a read-only instance.
Fixes an issue where running XA RECOVER on a read-only instance causes some transactions to be rolled back.
Fixes an issue where dates are missing in the enumeration of a time-based sharding function.
Fixes an execution error that occurs when the DISTINCT and ORDER BY clauses are used together.
Fixes the COLUMN 'S' NOT FOUND error in complex queries.
Fixes the COLUMN 'ID' IS AMBIGUOUS error in complex queries.
Fixes an error that occurs when you create an index that uses CURRENT_DATE as a column name.
Completes the help information for the SHOW HELP statement.

V5.3.9-15557587 (April 20, 2019)

CategoryDescription
New featuresSupports parallel DDL.
Supports refactored system variables, such as SELECT.
Supports materialization of subquery results.
Adds support for transaction statistics in SHOW STATUS.
Supports nesting of aggregates and expressions after SELECT.
Supports displaying MySQL physical execution plans in EXPLAIN.
Performance optimizationOptimizes the performance of IN and NOT_IN subqueries in some scenarios.
Fixed issuesFixes an issue where the default behavior of ALTER SEQUENCE is GROUP instead of AUTO_INCREMENT.
Fixes the "Convertor error by Unsupported" error.
Fixes the syntax error 'CHARSET=utf8 BLOCK_FORMAT=ENCRYPTED COMMENT' error.
Fixes an issue where adding an auto-increment primary key using ALTER TABLE incorrectly creates a SEQUENCE for a single table.
Fixes some errors that occur when transactions are enabled to improve transaction compatibility.
Fixes an error in a SELECT statement caused by the successful insertion of an invalid field of the YEAR type into a database.
Fixes an error that occurs in a CREATE TABLE … PARTITION BY RANGE COLUMNS statement.
Fixes an exception that occurs when you create a table with a secondary index if the index table rule contains dirty data with the same name.
Fixes an error caused by the default creation of system tables on volumetric read-only instances.
Fixes some issues with the DECIMAL function.

V5.3.8-15517870 (March 5, 2019)

CategoryDescription
New featuresAdds support for mapping rules.
Adds support for WHERE conditions in the SHOW FULL PROCESSLIST statement.
Adds support for the following cross-schema (database) operations:
  • Cross-schema queries.
  • Cross-schema BUSHY JOINs.
  • Cross-schema DML syntax.
  • Support for cross-schema XA transactions.
  • Cross-schema DAL syntax.
  • Cross-schema use of SEQUENCE.
  • Cross-schema access using USE DB_NAME.
Adds support for the PREPARE protocol.
Fixed issuesOptimizes the SHOW TABLES command to resolve slow execution issues.
Fixes a potential cross-shard error that occurs when the value of a time-based shard key is the NOW() function.
Fixes the "Unknown datatype name 'gbk'" error that occurs when you run CONVERT(column USING gbk).
Fixes an issue where the INTERVAL expression does not support fields, such as in DATE_FORMAT(DATE_ADD(MAX(table.column),interval table.column DAY),'%Y-%m-%d').
Fixes a null pointer exception (NPE) that occurs when an INSERT statement contains a CASE WHEN clause.
Fixes an issue where the Hint: Allow full table scan = false setting does not take effect.

V5.3.7-15460044 (December 28, 2018)

CategoryDescription
New featuresSupports adjusting the JOIN pushdown order to optimize JOIN performance.
Supports adjusting the in-memory JOIN order based on the number of table rows.
Adds support for logical database locks.
Supports displaying the PREPARE and EXECUTE syntax.
Adds HINT support for the inter-table parallel execution mode.
Performance optimizationsImproves the error messages for logon failures caused by whitelist or password issues.
Optimizes multi-table JOINs to resolve performance issues caused by the generation of redundant subqueries in physical SQL.
Optimizes the executor to prevent it from running SET NAMES utf8mb4 multiple times when a connection is obtained.
Optimizes the performance of SHOW VARIABLES.
Fixed issuesFixes an issue where SQL statements are not fully displayed in SHOW FULL PROCESSLIST.
Fixes an error that occurs with SQL functions, such as ~1.
Fixes an error that occurs when you import a database and an error that occurs when you create a table after the import.
Fixes an error that occurs in a multi-table JOIN, such as (index (16) must be less than size (16)).
Fixes an issue where the processing of LIMIT by SCAN HINT does not meet expectations.

V5.3.6-15447056 (December 13, 2018)

CategoryDescription
Fixed issuesFixes an error that occurs when a user-created account queries information_schema, which affects the use of DMS and some GUI tools.
Fixes a performance degradation issue in some queries caused by inaccurate pushdown of Filter-Join execution plans.
Fixes an issue where a redundant subquery is included in the physical SQL statement generated in some scenarios, which prevents MySQL indexes from being used.
Fixes an issue where UNION optimization causes abnormal physical SQL statements to be generated in some scenarios.
Fixes an alias issue in the physical SQL statement generated for a correlated subquery.
Fixes an issue where an abnormal query result is returned after LIMIT is pushed down through a JOIN in some scenarios.
Fixes an issue where the SHOW CREATE TABLE statement does not display the table sharding algorithm.
Fixes an issue where a string column cannot be specified in UNI_HASH when you create a table.

V5.3.6-15439241 (November 29, 2018)

CategoryDescription
New featuresSupports granting a single account permissions on multiple databases and using USE DATABASE to switch databases and run SQL statements.
Adds support for all column subqueries in SELECT statements and subqueries in WHERE clauses.
Adds support for the Apply subquery operator, which lets you run nested subqueries that cannot be converted to JOINs.
Adds support for the HashBKAJoin operator.
Supports pushdown execution of multi-table UPDATE and DELETE statements.
Supports pushdown execution of geographic information functions.
Supports column order detection for heterogeneous index tables to prevent DDL from causing index table synchronization exceptions.
Supports IPv6.
Performance optimizationsOptimizes the system to prohibit the use of the BINARY type as a shard key. This change prevents query exceptions caused by automatic zero-padding during data insertion.
Optimizes the parser performance for end-to-end stress testing.
Fixed issuesFixes a query exception caused by an incorrect SQL statement being sent when NODE HINT and LIMIT are used together.
Fixes a low LEFT JOIN performance issue caused by the failure to push down some conditions.
Fixes an execution exception that occurs when BKA JOIN and LIMIT are used together in some scenarios.
Fixes an out-of-bounds issue that occurs when you set the maximum value for a Simple Sequence.
Fixes an error that occurs in a CREATE TABLE statement for an existing table when the IF NOT EXISTS keyword is used.

V5.3.5-15423348 (November 20, 2018)

CategoryDescription
Fixed issuesFixes an issue where the selection of a concurrent policy causes abnormal sorting results in some scenarios.
Fixes a compatibility issue between NODE HINT and LIMIT that causes abnormal results in some scenarios.
Fixes an issue where a change in the transaction policy cleanup mechanism causes DTS tasks to become abnormal in some scenarios.
Fixes an execution exception that occurs in a BATCH INSERT statement in some scenarios.
Fixes an error that occurs when a BATCH INSERT statement processes special escape characters.

V5.3.5-15409153 (October 30, 2018)

CategoryDescription
New featuresAdds full support for the GROUP_CONCAT syntax.
Adds support for the STR_HASH hashing function, which is compatible with all HASH functions in PolarDB-X 1.0 V5.2.
Supports the use of aliases in UPDATE statements.
Supports specifying table names in the ON DUPLICATE KEY UPDATE syntax.
Supports pushdown execution of JSON-related functions.
Supports the CONVERT(X, SIGNED) syntax.
Supports the JOIN USING syntax.
OptimizationsSignificantly reduces the memory usage of BATCH INSERT operations to improve the stability of data imports.
Optimizes the INSERT concurrency policy to resolve the issue of high backend connection usage.
Fixed issuesFixes an error about duplicate column names in a JOIN when the physical SQL statement contains a subquery.
Fixes some minor issues related to OUTLINE.
Fixes an issue where the return value of the TO_DAYS function is inconsistent with MySQL during a leap year.
Fixes an error that occurs when a keyword is used as the column name for a primary key or shard key during table creation.
Fixes an error that occurs in a special INSERT SELECT syntax.
Fixes an execution plan exception that occurs when a function is included in both the ORDER BY and GROUP BY clauses.
Fixes an issue where a BIGINT column is represented as DECIMAL in the result set.
Fixes an error that occurs in a SHOW FULL TABLES FROM LIKE statement.
Fixes an error that occurs when a SHOW DATABASE statement contains a line feed.

V5.3.4-15378085 (September 20, 2018)

CategoryDescription
New featuresSupports synchronous updates of broadcast tables. This feature is based on multi-write XA transactions and does not require asynchronous data synchronization.
Supports various dynamic parameters, such as the global transaction policy, which can be set in the console.
Uses XA transactions as the default transaction policy.
Supports SET @@SESSION.BINLOG_ROWS_QUERY_LOG_EVENTS.
Fixed issuesFixes an issue with the BIT type.
Fixes a JOIN query error caused by BKAJoin+UNION optimization.
Fixes a type mismatch issue in scalar subqueries.
Fixes an issue where the CURRENT() method is not implemented in the SEMI and ANTI executors.
Fixes an issue where a special syntax of CURRENT_TIMESTAMP is not supported.

V5.3.3-1670435 (August 15, 2018)

CategoryDescription
New featuresAdds support for XA transactions in PolarDB-X 1.0.
Supports unitized Group Sequence.
Performance optimizationsOptimizes the handling of MysqlDump compatibility issues by adding support for MySQL executable HINTs.
Optimizes the handling of MysqlDump compatibility issues by adding support for the MySQL single-line comment "--".
Fixed issuesFixes an NPE caused by the incorrect setting of a ThreadLocal variable when a HINT exists.
Fixes a full table scan issue caused by the failure to pull up subquery conditions in some scenarios.
Fixes a full table scan issue caused by the failure to extract shard key conditions from a subquery.
Fixes a full table scan issue that occurs because the BETWEEN … AND clause is not considered during physical shard calculation.
Fixes an issue where a `FOR UPDATE` SQL statement that can be directly pushed down to RDS may be routed to a secondary database.
Fixes an issue where the return value of SELECT @@session.tx_read_only is inconsistent with MySQL.
Fixes an issue where the calculation result of an arithmetic expression is incorrect in some cases.
Fixes an error that occurs when processing the value of the 0000-00-00 00:00:00 timestamp.
Fixes an out-of-memory (OOM) issue that may occur in a NESTLOOP JOIN operation when the data volume is large.
Fixes an NPE that occurs in some query scenarios when a DRDS HINT exists.
Fixes an error caused by variable passing between PolarDB-X 1.0 and RDS when a transaction is enabled.
Fixes a parser error that occurs when multiple PolarDB-X 1.0 HINTs are added before an SQL statement.
Fixes an issue where the output of a physical slow SQL statement contains an incorrect physical table name.
Fixes an issue where a coroutine excessively occupies CPU and affects I/O responses due to a large amount of operator computation.
Fixes an issue where the UNIX_TIMESTAMP() function is not supported without parameters and returns NULL when parameters are provided.
Fixes an incorrect routing issue of UNI_HASH for bigint unsigned and int unsigned.
Fixes an issue where an INSERT statement with a HINT that specifies a physical database and table reports that the table does not exist.
Fixes an issue where Group Sequence is incorrectly used when you create a table in a single database.
Fixes a connection leak issue caused by a "because null" error. The error is triggered by an abnormal fail-fast recovery of the Druid connection pool in some special cases.
Fixes a physical SQL query error caused by the parameterization of the LIMIT clause in an SQL statement. The parameterization of the LIMIT clause has been canceled.
Optimizes the rules for accurately selecting the BKAJoin algorithm to resolve the issue of high memory usage in non-pushdownable JOINs.
Fixes an error that occurs in a sent SQL statement when the INSERT VALUE contains an escape character.
Fixes a sharding calculation error caused by incomplete condition derivation in a LEFT JOIN.
Fixes a StackOverflow issue in a BKAJoin operation when the driver table is large.
Fixes a connection leak issue caused by the failure to close a connection promptly after a Union query is optimized.
Fixes a case mismatch issue with a shard key caused by a special sharding rule.
Fixes a memory leak issue related to type-related metadata.
Fixes an issue where the parameterization cache is not cleared promptly after an SQL statement is run, which occupies memory.
Fixes an error caused by an incorrect pushdown of the execution plan for a cross-database JOIN query that involves the UNI_HASH, RANGE_HASH, or RIGHT_SHIFT sharding function.
Fixes an error that occurs in a DMS or DTS operation because the database name metadata returned by Java Database Connectivity (JDBC) after an SQL statement is run is incorrect.
Fixes an error that occurs if the `set drds_transaction_policy` statement for XA transaction verification is not placed before the start of the transaction.

V5.3.2-1645122 (July 5, 2018)

CategoryDescription
Fixed issuesFixes an issue where SQL statements with an Outline fail to take effect in high-concurrency scenarios, which results in unexpected execution plans.
Fixes an error in obtaining a physical connection. The error occurs when a SELECT statement for a single physical database in a transaction uses a parallel query policy.

V5.3.2-1623342 (May 25, 2018)

CategoryDescription
New featuresImproves native transaction support for XA transactions.
Adds support for V5.2 and V5.3 HINTs in INSERT, UPDATE, and DELETE statements.
Fixed issuesFixes an issue where the table name in the result set metadata is incorrect.
Fixes an issue where the precision and range of some data types are abnormal.
Fixes an error that occurs when you use a Direct HINT and specify only some logical table names.
The SHOW NODE command reports only the read request ratio.
Fixes a query exception that occurs when a keyword is used as a table name in a database sharding without table sharding scenario.
Fixes a multi-UNION query exception.
Fixes an issue where INSERT last_insert_id returns 0 for a single table.