Versions

openGauss 5.0.0 is the third LTS version of openGauss, and its lifecycle is three years. This document applies only to the lite edition.

5.0.0 is compatible with the earlier versions. Main functions are as follows:

• Inherited functions:

  • SQL standard syntax, UPSERT, data type, XML type, table, temporary table, global temporary table, foreign table, view, materialized view, index, foreign key, generalized inverted index (GIN), sequence, function, trigger, ROWNUM, MEDIAN aggregate function, JSONB data type, GB18030 character set, UPSERT subquery, and row-store execution to vectorized execution.

  • Stored procedure, commit/rollback in stored procedure, omission of parameter parentheses () from the stored procedure or function calling, stored procedure debugging, and autonomous transaction.

  • Security features such as authentication, permission management, network communication security, database audit, SM algorithm, and built-in role and permission management.

  • HA functions such as primary/standby two-node cluster, cascaded standby node, logical replication, ultimate RTO, standby node scale-out, and Paxos-based distributed consensus framework (DCF).

  • Range partitioning, global partitioned indexes, list partitioning, hash partitioning, and interval partitioning (automatic partitioning based on range partitioning).

  • Full physical backup, logical backup, standby node backup, incremental backup and restoration, and point-in-time recovery (PITR).

  • NUMA-aware high-performance optimization, parallel query, Global SysCache, and automatic eviction of unique SQL statements.

  • Storage features such as delayed standby node replay, logical replication on standby nodes, Xlog archiving on standby nodes, parallel logical decoding, hash indexes, and row-store table compression.

  • Load balancing and read/write isolation on the JDBC client, CMake script compilation, container-based deployment, IPv6 protocol, and postgis plug-in.

  • Toolchain: development tool DataStudio

  • Middleware: ShardingSphere, Kubernetes, and openLookeng.

  • Basic functions: standard SQL syntax, data types, tables (including temporary tables, global temporary tables, and foreign tables), views, materialized views, foreign keys, indexes (including B-tree indexes, GIN indexes, and hash indexes), sequences, functions, triggers, MEDIAN aggregate function, ROWNUM, UPSERT, JSONB data type, and GB18030 character set.

  • Enhanced SQL engine: range partitioning, global partitioned index, list partitioning, hash partitioning, automatic extended partitioning based on range partitioning, row-store execution to vectorized execution, autonomous transaction, parallel query, Global SysCache, IPv6 protocol, and PostGIS extensions.

  • Enhanced storage engine: delayed standby node replay, logical replication on standby nodes, parallel logical decoding, gray upgrade, rolling upgrade, hash index, unique primary key constraint for column-store tables, publication and subscription, and NUMA-aware high-performance optimization.

  • Stored procedure, commit/rollback in a stored procedure, omission of parameter parentheses () from the stored procedure or function calling, and stored procedure debugging.

  • Security functions: authentication, permission management, network communication security, database audit, Chinese national cryptographic algorithm, built-in role and permission management, and ANY permission management.

  • High availability: primary/standby two-node cluster, cascaded standby node, logical replication, ultimate RTO, standby node scale-out, and Paxos-based distributed consensus framework (DCF),

  • Backup and restoration: Full physical backup, logical backup, standby node backup, incremental backup and restoration, and point-in-time recovery (PITR).

  • O&M capabilities: database running metrics in WDRs, diagnosis view for slow SQL statements on the standby node, and automatic eviction of unique SQL statements.

  • JDBC client load is balanced and read and write are isolated.

  • Toolchain: development tool DataStudio, MySQL full migration tool, incremental migration tool, reverse migration tool, and data check tool.

  • Middleware: ShardingSphere and openLookeng.

  • Peripheral ecosystem: dblink, supporting openEuler, CentOS, and FusionOS.

  • Others: CMake script compilation, container-based deployment, and Kubernetes.

• New functions:

  • High performance: The performance of basic operators is improved. (For details, see the feature description.)
  • High security: user-level audit.
  • Enterprise-level feature: event trigger.
  • Application development interface: Embedded C preprocessor
  • Application development interface: enhanced JDBC capabilities.
  • Middleware: ShardingSphere supports read/write isolation in multiple modes and cross-database query.
  • Middleware: 32-node networking performance of openGauss+ShardingSphere > 21 million tpmC
  • Toolchain: full-lifecycle data production tool DataKit
  • Toolchain: MySQL migration tool gs_rep_portal
  • Toolchain: MySQL incremental and reverse migration tool gs_replicate
  • Toolchain: gs_datacheck based on the Merkle tree
  • Compatibilities: Compatible with most common MySQL syntax, including data types, operators, system functions, DDL, DML, DCL, and PL/SQL, implementing application migration without modification.

• Fixed defects:

  • I6NVAB: An error is reported during rollback and forcible rollback after gray upgrade from 3.0.3 to 5.0.0.
  • I6NK8U: When a publication/subscription name is changed to an existing name, the error message “ERROR: unsupported object class xxx” is displayed.
  • I6NK9V: The pg_os_threads view frequently allocates memory. As a result, the access efficiency is low.
  • I6LRP5: After a session exits or a dblink is deleted, the connection can still be queried in the remotely connected database.
  • I6JEG9: After a column is added to an column-store table and the transaction is rolled back, the system breaks down or an error is reported when the transaction is redone.
  • I6N2QD: When ESCAPE is used in SIMILAR TO, if the escape character is Chinese, an error is reported. If the LIKE escape character is Chinese, the operation is successful.
  • I6MOCS: In the B-compatible database, if the partition key of a level-2 partitioned table is the year function, an unknown error is reported when pg_get_tabledef is used after the ADD PARTITION operation is performed.
  • I6LMTU: In B-compatible mode, objects in backquotes are not supported.
  • I6K2BZ: Chameleon migration fails due to double quotation marks.
  • I6KIWQ: During Debezium+kafka incremental migration, the timestamp after the migration is incorrect.
  • I6I640: When the data check tool is used to check data in 150,000 tables, the check process is suspended.