Release: 0.24.5

Since 0.24.4, there is a new experimental feature in Zeebe which checks for inconsistencies when replaying the log - that is, when the workflow engine is restarting, it will attempt to verify that it rebuilt the in-memory state correctly. This was primarily built to assert that upgrades were successful or not. Zeebe 0.24.5 introduces a feature flag which controls whether or not this feature is enabled. As we've found a few bugs in there, we recommend you upgrade to 0.24.5 as soon as possible, which will allow you to turn on this feature only during upgrades; once you've verified that the upgrade was successful, you can then restart your nodes without the flag. In the meantime we will keep on polishing this feature such that it can be used at all times.

Enhancements

Broker

• Control inconsistency detection with an experimental feature flag (#5759)

Bug Fixes

Broker

• Re-processing detects inconsistency due to Deployment distribution (#5568)
• Failed to take snapshot. Expected to find an indexed entry for given snapshot position, but found no matching indexed entry which contains this position (#5294)

Merged Pull Requests

• [BACKPORT 0.24] Add experimental configuration flag to control inconsistency detection (#5768)
• [Backport 0.24] Ignore deployment records in reprocessing issue detection (#5583)
• [Backport 0.24] Fix flaky CancelWorkflowInstanceConcurrentlyTest again (#5553)
• [Backport 0.24] Don't update exporter position to smaller value (#5527)