Broker Release Notes
Last updated:2026-01-20 17:36:15
This document provides the records of TDMQ for CKafka (CKafka) kernel minor version upgrades and changes.
Version 3.2.3
Version | Feature | Remarks |
3.2.3_1.0.10 | Aligned with version 2.8 and supplemented commits. Enabled dynamic adjustment of produceId/transactionId expiration time. Changed the error code of broker failures caused by full disk to BrokerNotAvailable. Enabled dynamic change of offsets.retention.minutes. Cleared the released connections in TCP connectionIpMap, and enabled periodic scanning. Added the rules for consumer group name verification. Enabled the display of subscription information for consumer groups subscribing in the Consumer Assign mode if offsets are committed. | - |
3.2.3_1.0.7 | Added acknowledgment node-level and instance-level downgrade capabilities. Enhanced the server fallback capability of recovery in case of transaction/production failure. Optimized ACL logs of IP range validation failures. Optimized the broker processing logic of internal communications permissions. | - |
3.2.3_1.0.5 | Improved the notification scenario coverage of the Event Center. Improved the ZooKeeper design to prevent unexpected broker shutdowns caused by session timeout. Improved the transaction mechanism to prevent the log cleanup thread failure in the state of empty transactions. | - |
3.2.3_1.0.4 | Improved the accuracy of production and consumption traffic throttling. | - |
Version 2.8.1
Version | Feature | Remarks |
2.8.1_1.1.11 | New architecture features Added the layered two-phase metadata optimization. Added the layered cold read cache optimization. Added the capability of hitless layered Cloud Object Storage (COS) data migration to another bucket. Feature optimization Optimized the performance of batch deletion of layered cold data. Optimized the performance of the RemoteIndex loading lock. Added and improved critical path metrics and tools for layered modules. Corrected the issue of COS data read amplification caused by cold read range requests. Stability enhancement Added metadata parsing and query tools. Upgraded COS SDK to the stable version 5.6.244.2. Unthrottled internal layered topic read and write. Adjusted the limit on the number of remote index cache files. | - |
2.8.1_1.1.10 | Log optimization Community issue: Merged KAFKA-13636. Enabled dynamic adjustment of produceId/transactionId expiration time. Changed the error code of broker failures caused by full disk to BrokerNotAvailable. Introduced multi-tenant features. Cleared the released connections in TCP connectionIpMap, and enabled periodic scanning. Added the rules for consumer group name verification. Enabled the display of subscription information for consumer groups subscribing in the Consumer Assign mode if offsets are committed. Enabled dynamic change of offsets.retention.minutes. | - |
2.8.1_1.1.8 | Repaired dirty consumer data in elastic storage instances. Fixed the issue of excessive client error logs. Fixed the issue of excessive CPU consumption caused by internal port parsing from CSV files during request processing. | - |
2.8.1_1.1.7 | Added a method for checking the completion of remote topic consumption in AdminClient. Added the metrics of MetadataCatchUpLag, MetadataCatchUpReadOffset, and cache hit rate. Avoided cache clearing before consumption completion, and increased the timeout duration for remote data read. | - |
2.8.1_1.1.6 | Added acknowledgment node-level and instance-level downgrade capabilities. Enhanced the server fallback capability of recovery in case of transaction/production failure. Merged the transaction query tool of the open-source edition. Optimized ACL logs of IP range validation failures. Fixed the issue of unbalanced synchronization caused by dynamic adjustment of the number of replica synchronization threads. Optimized the broker processing logic of internal communications permissions. | - |
2.8.1_1.1.3 | Improved the notification scenario coverage of the Event Center. Added a pay-as-you-go storage product offering. Improved the ZooKeeper connection design to prevent unexpected broker shutdowns caused by session timeout. Improved the leader election mechanism to prevent a coordinator from being elected or uninstalled by mistake. Improved the leader processing logic to prevent the stop of the follower synchronization thread before offsetOutOfRange is correctly processed. | - |
2.8.1_1.1.2 | Improved the accuracy of production and consumption traffic throttling. | - |
2.8.1_1.0.9 | Fixed the issue of consumption traffic throttling accuracy. | - |
2.8.1_1.0.8 | Enabled topic-level traffic throttling. | - |
2.8.1_1.0.7 | Enabled SCRAM 512 authentication. | - |
2.8.1_1.0.6 | Fixed KAFKA-12835: Topic IDs can be mismatched on brokers (following interbroker protocol version updates). | - |
2.8.1_1.0.5 | - | |
2.8.1_1.0.4 | Added the switch for automatic creation of consumer groups. Enabled SCRAM authentication. Added the feature for viewing the number of connections for producers. | - |
2.8.1_1.0.3 | Disabled monitoring reporting#. | - |
Version 2.4.2
Version | Feature | Remarks |
2.4.2_1.2.7 | Optimized logs. Community issue: Merged KAFKA-10706. Enabled dynamic adjustment of produceId/transactionId expiration time. Changed the error code of broker failures caused by full disk to BrokerNotAvailable. Used the default values of CreateTopic PartitionNum and ReplicaNum defined by the user in the console if the two parameters are left blank. Cleared the released connections in TCP connectionIpMap, and enabled periodic scanning. Added the rules for consumer group name verification. Enabled the display of subscription information for consumer groups subscribing in the Consumer Assign mode if offsets are committed. Enabled dynamic change of offsets.retention.minutes. Community issue: Merged KAFKA-7188 to solve the issue of slow connection when DNS fails. | - |
2.4.2_1.2.4 | Added acknowledgment node-level and instance-level downgrade capabilities. Enhanced the server fallback capability of recovery in case of transaction/production failure. Optimized ACL logs of IP range validation failures. Fixed the issue of unbalanced synchronization caused by dynamic adjustment of the number of replica synchronization threads. Optimized the broker processing logic of internal communications permissions. | - |
2.4.2_1.2.2 | Improved the notification scenario coverage of the Event Center. Broker traffic (not aggregation of partition data and excluding replicas). | - |
2.4.2_1.2.1 | Improved the accuracy of production and consumption traffic throttling. | - |
2.12-2.4.2_1.1.2 | Merged a community bug fix in Consumer offset reset after new segment rolling. For details, see KAFKA-9543. | - |
2.12-2.4.2_1.1.1 | For determining whether a request is from a protected port, store listenerName and listenersSize in variables first to avoid high CPU consumption caused by excessive requests. Enabled topic traffic throttling. | - |
2.12-2.4.2_1.0.8 | Added the limit for automatic creation of consumer groups. Enabled SCRAM authentication. | - |
2.12-2.4.2_1.0.7 | Enabled token authentication. | - |
2.12-2.4.2_1.0.6 | Allowed output of production time consumption logs. Disabled monitoring reporting#. | - |
2.12-2.4.2_1.0.5 | Enabled statistics of producer connections. | - |
Version 1.1.1 (discontinued)
End of marketing (EOM) on March 6, 2024.
Version | Feature | Remarks |
1.1.1_1.1.14 | Fixed the issue of the number of groups exceeding the limit. Fixed the NPE issue in superuser authentication. Enhanced the server fallback capability of recovery in case of transaction/production failure. Optimized ACL logs of IP range validation failures. Fixed the issue of unbalanced synchronization caused by dynamic adjustment of the number of replica synchronization threads. | - |
1.1.1_1.1.12 | Improved the notification scenario coverage of the Event Center. Improved the broker shutdown design to prevent restart failures caused by an unclean shutdown. Improved dynamic modification of topic configurations to prevent the issue of original configurations being discarded during the modification. Improved the network connection design to prevent connection leaks. | - |
1.1.1_1.1.11 | Improved the accuracy of production and consumption traffic throttling. Fixed the issue where the displayed number of connections is 0. Optimized the reuse capability of connections created. | - |
2.11-1.1.1_1.1.8 | Allowed SCRAM VPC access to new Standard Edition instances. | - |
2.11-1.1.1_1.1.6 | Adjusted JVM parameters for container runtime. | - |
2.11-1.1.1_1.1.4 | Enabled topic traffic throttling. | - |
2.11-1.1.1_1.1.3 | Fixed the issue of ulimit configuration. | - |
2.11-1.1.1_1.1.2 | Added client IP logs in getMetadata. | - |
2.11-1.1.1_1.1.1 | Graceful downgrade for ack=all. | 1. This version is independent of branch 1.1.1_1.1.1-ackDowngrade. 2. 1.1.1-publish is displayed if the code is not merged. 3. The ticket submitted by customers is escalated. |
2.11-1.1.1_1.1.0 | Optimized the getMetadata lock performance. | - |
2.11-1.1.1_1.0.13 | Enabled token authentication. | - |
2.11-1.1.1_1.0.12 | Allowed output of production time consumption logs. | - |
2.11-1.1.1_1.0.11 | Enabled statistics of producer connections. | - |
Version 0.10.2 (discontinued)
End of marketing (EOM) on March 6, 2024.
Version | Feature | Remarks |
0.10.2.1_1.2.11 | Fixed the issue of groups exceeding the quantity limit. | - |
0.10.2.1_1.2.9 | Fixed the issue of startup failure caused by data exceptions in /kafka-acl. | - |
0.10.2.1_1.2.8 | Enabled statistics of producer connections. | - |
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback