emqx Files

Enhancements

• #11469 Added support for specifying username in Redis authentication.

• #11487 The bcrypt work factor is limited to the range 5-10, because higher values consume too much CPU resources. Bcrypt library is updated to allow parallel hash evaluation.

• #11496 Disabled the Erlang VM Prometheus exporter by default to improve performance and security.

• #11497 Enhanced broker metrics collection and export by adding new metrics for messages, overload protection, authorization, authentication, and improving naming consistency for OpenTelemetry.

• #11490 Added fast error handling for undefined passwords in various authentication backends. This improves the consistency and user-friendliness of the authentication process.

• #11532 Improve some error reasons when parsing invalid packets.

Bug Fixes

• #11466 Fixed a crash that occurred when setting the ssl_options.ciphers configuration option to an empty string ("").

• #11480 Improves the error handling and testing of SQL functions in the rule engine when rule functions receive bad arguments.

• #11493 Examples and documentation for /api/v5/publish bad request response have been fixed. Previously the documentation example said that the bad request response could return a list in the body which was not actually the case.

• #11506 Avoid downloading a trace log file if it is empty.

After this fix, GET /api/v5/trace/clientempty/download returns 404 {"code":"NOT_FOUND","message":"Trace is empty"} If no events matching the trace condition occurred.

• #11520 Fixed issue where packets_connack_sent metric was not incremented on CONNACK packets sent with non-zero ack_flag.

• #11522 Improved error message for rule engine schema registry when schema name exceeds permissible length.

• #11523 Corrected a misleading prompt when specifying invalid certificates/keys for the /configs API.

• #11531 Fixed issue where authorization cache cleaning cli was not working properly for specific client ID.

• #11534 Fixed the increment on data bridge statistics when the bridge is unhealthy. Now, messages sent to unhealthy bridges are counted as dropped messages.

• #11540 Improved HTTP response when attempting to create a bridge with an invalid name.

• #11548 Fixed an issue that prevented the plugin order from being updated across the entire cluster.

• #11564 Fix cluster partition autoheal functionality. Implement autohealing for the clusters that split into multiple partitions.