TDengine 3.4.1.6 Release Notes
Enhancements
- When creating taosX tasks in taos-explorer, users are automatically redirected to the XNODE creation page if no XNODE is available
- In Explorer DataIn, the default empty tag column is automatically removed when a supertable data column is moved to the tag column
- Improved cross-origin security in taos-explorer
- Added an XNODE management page to taos-explorer
- Optimized communication logging between xnoded and taosX
- Improved SQLAlchemy compatibility in taos-ws-py
- Added support for importing and exporting taosX configuration files through taos-explorer
- Added automatic reconnection support for the AVEVA Historian data source in taosX after disconnection
- Added support in the Rust connector for subscribing to batch tag update messages
- Added DECIMAL and BLOB support to parameter binding in the Rust connector
- Added a connector security development guide and security examples
- Added the -H option to taos shell to display BINARY strings containing non-printable characters in hexadecimal format (0x...), disabled by default
- Added daylight saving time conversion support to the TO_ISO8601 function
- Added minute-level timezone offset support for the TZH format in the to_char and to_timestamp functions
- In STMT2 write mode, child-table metadata changes no longer return the TSDB_CODE_NEED_RETRY error
- Interval windows now support partitioning based on the timezone carried by the client
- Optimized the output format of show table distributed and added the INFORMATION_SCHEMA.INS_DISK_USAGE system table for querying table data distribution
- Optimized the truncation logic when the last WAL file is corrupted
- Tables used by subscriptions can now be dropped. After dropping and recreating the table, running reload topic restores the subscription
- Added the syncAssignedCheckAppliedGap parameter. In dual-replica mode, when a node is assigned as leader, this parameter controls the threshold gap between the follower's applied index and commit index to prevent the follower from entering the restoring state before catching up
- Added DECIMAL column support for virtual tables
Fixes
- Fixed slow loading of the Explorer DataIn tasks page when many tasks exist
- Fixed inaccurate SQL execution times returned by taos-explorer
- Fixed the issue where filter conditions did not take effect when configuring taosX tasks in taos-explorer
- Fixed backup point generation errors in taosX when no data is available
- Fixed an internal logic error that caused OPC access failures on Windows when taosX bound to port 6051
- Fixed the issue where the authentication parameter input box was not displayed when Custom Authentication was selected for the Pulsar data source in taosX
- Fixed the issue where the Pulsar data source in taosX remained in the creating state because data splitting is unsupported
- Fixed incorrect auto-increment behavior when taosX generated timestamp column data with composite primary keys for Kafka data sources
- Fixed the issue where taosX did not display an error and could not recover or retry when fetching sample data for SparkPlugB
- Fixed the intermittent failure to stop xnoded child processes on Windows
- Fixed the issue where taosdump returned empty data without throwing an exception when the engine service became unavailable during WebSocket reads
- Fixed significant performance degradation when taosdump exported large-volume subtables
- Fixed incorrect path handling when generating keys with the taosk tool
- Fixed the issue where the queryTableNotExistAsEmpty parameter did not take effect when accessed through taosAdapter
- Fixed errors when creating streams in nanosecond-precision databases if fixed values were used for TAG columns in stream output tables
- Fixed conflicts between expired data deletion (trim) and shared-storage migration operations
- Fixed a potential issue where the mnode message queue could become saturated during data compact operations if some nodes stayed offline for an extended period, causing connection exceptions
- Fixed failures when writing monitoring information to the log database
- Fixed a latent risk caused by inconsistent column counts between virtual subtables and virtual supertables in the schema
- Fixed crashes caused by executing certain functions in queries without a FROM clause
- Fixed crashes caused by type-conversion expressions such as not (cast(null as varbinary) between now() and ts) in queries
- Fixed errors when writing stream computing results to an existing regular table as the result table
- Fixed an OOM issue caused by incorrect memory release logic when using %%tbname in stream queries under large-write workloads
- Fixed high memory usage in count_window when reading historical data in stream computing
- Fixed OOM crashes caused by an uncontrolled upper limit on the number of sliding windows during WAL replay in stream computing
- Fixed crashes in outer stream queries when diff() and notify were used together
- Fixed crashes and incorrect results in the group_concat function
- Fixed inconsistencies between the execution plan and actual execution when CASE WHEN referenced the primary key as a filter condition
- Fixed a memory leak on taos_cleanup exit
- Made the output of USE DATABASE on macOS consistent with Linux; it now uniformly displays Database changed.
- Fixed the issue where the Windows installer could not shut down processes gracefully