Release Notes
Announcements
Version Information | Core Issue | Fix Description |
Redisson < 3.21.0 | In Redis cluster mode, when backend nodes change (such as failover or scaling nodes), the Redisson client fails to properly release old node connections, which may cause connection leaks in the connection pool. | |
Redisson < 3.17.5 | When replica node changes (such as HA switchover, scaling, or configuration change) occur in a Redis cluster, incorrect replica refresh may cause Redisson lock acquisition exceptions. | This issue has been resolved in Redisson 3.17.5. For details, see Error while Attempting to Set Lock. |
Redisson < 3.18.1 | When connection interruptions occur due to network exceptions or server crashes, if the business system is using a distributed lock, all connections that acquired the lock will remain occupied and unable to be released. The client continuously throws java.util.concurrent.CancellationException errors, eventually leading to the entire distributed lock feature becoming completely unavailable. | |
Redisson > 3.44.0 | In standalone mode, when Redisson connects to Tencent Cloud Redis and calls the tryLock method, its recently added lock renewal mechanism sends Lua scripts based on the thread count, causing cross-slot operation errors. Note: In Redisson cluster mode, this issue can be resolved. Still, you need to verify whether there is any exception due to conflicts between the cluster discovery mechanism and the VIP architecture during the connection establishment phase, and whether VIP switchover during instance configuration changes (like scaling) affects client stability. | To resolve this issue in standalone mode, explicitly set the lockWatchdogBatchSize parameter to 1. This configuration will force Redisson's lock batch renewal mechanism to execute Lua scripts in single-thread serialization, ensuring each renewal operation processes only one key, thereby avoiding cross-slot errors.
|
フィードバック