Updated on 2024-04-01 GMT+08:00

Kernel Version Notes

This section describes the kernel version updates of DDM.

Table 1 Kernel version description

Version

Description

3.1.2.0

New features

  • Improve the troubleshooting capability of the connection pool, and add the fail-fast feature (disabled by default) for data nodes.
  • Improve data verification of shard configuration.
  • Optimize memory configuration.
  • Support the ANALYZE TABLE statement.
  • Rebuild global secondary index for whitelisted users.

3.1.1

New features

  • Support global secondary index for whitelisted users.
  • Support SSL encrypted connection.

3.1.0

New features

  • Add filter criteria for Show Processlist and Kill Sessionid.
  • Add keywords related to MySQL 8.0 Online DDL.

Resolved issues

  • Fixed the issue that the millisecond value is lost when UPDATE statements are executed using the DATE_SUB function.
  • Optimized the display of field aliases in some complex queries.
  • Fixed the issue that new tables cannot be used when rename operations are performed after other DDL operations.
  • Fixed the error occurred when sequences are concurrently inserted.

3.0.9

New features

  • Support the query for auto-increment progress of a sequence.
  • Support the query for a record whose size exceeds 16 MB.
  • Support MariaDB Connector/J.
  • Optimize transaction splitting. If there are no updates in a transaction, requests are routed to the primary instance or a read replica based on read/write weights.

3.0.8

New features

  • Support route calculation of ROW expressions.
  • Support table recycle bins.
  • Support metadata backup and restoration.

Resolved issues

  • Optimized the DDM MetaDB connection pool.
  • Optimized the DDM process monitoring mechanism.
  • Optimized the display of Show Processlist.
  • Optimized the process of Reload for obtaining metadata.
  • Optimized the process for reporting an error when a transaction rollback occurs during incremental playback.

3.0.6

New features

  • Allow users to connect to metadata using connection pools during shard change.

Resolved issues

  • Fixed the issue that the end status of a DDL command being executed cannot be recorded when Ctrl+C is pressed.
  • Optimized the high CPU usage when a process is killed during big data query on the client.

3.0.4

Resolved issues

  • Fixed the error of read/write splitting in some abnormal scenarios.
  • Optimized the execution of the GROUP_CONCAT function when there are a large amount of data.