This is an early access release for JBOD in KRaft (KIP-858), and it is not recommended for use in production environments. Instead, we advise users to create new clusters with version 3.7.0-IV2 and test the feature there. Alternatively, you can try the JBOD storage feature on existing non-production clusters upgrading them to version 3.7.0-IV2 or upgrading existing non-production cluster running JBOD in Zk to version 3.7.0-IV2 and then migrating them to KRaft.
While the early access release of JBOD in KRaft offers the opportunity to try out this new feature, it is important to be aware of the following limitations:
- At least two known bugs can cause log directory failures to re-create partitions in another logdir : When a single log directory fails and then is subsequently brought back on line, the restart of the broker will fail because the same partitions will occur in two different directories. This can also happen as a combination of intra-broker replica movement and log directory failure. These issues are tracked in KAFKA-16234 - Getting issue details... STATUS and KAFKA-16082 - Getting issue details... STATUS .
- A race condition while promoting a future replica, as a result of intra-broker replica reassignment, can lead to unavailability. KAFKA-16297 - Getting issue details... STATUS
For the latest information regarding known issues, their resolutions, and possible workarounds, please visit the parent tracker for KRaft JBOD follo-ups and improvements at KAFKA-16061 - Getting issue details... STATUS . We are committed to addressing these issues and providing a reliable JBOD in KRaft experience, and your feedback is incredibly valuable in helping us improve the feature.