[CASSANDRA-20116] Avoid prepared statement invalidation race when committing schema changes

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 5.1-alpha1
• Component/s: Transactional Cluster Metadata
• Bug Category: Correctness - Transient Incorrect Response
• Severity: Normal
• Complexity: Low Hanging Fruit
• Discovered By: Adhoc Test
• Platform: All
• Since Version: 5.1-alpha1

Source Control Link

https://github.com/apache/cassandra/commit/9bf1680b1f1c016717381aa59367950fa770f57f

Test and Documentation Plan

new and existing tests

• Assignee: Marcus Eriksson
• Reporter: Marcus Eriksson

Authors

Marcus Eriksson

Reviewers

Sam Tunnicliffe

• Created: 27/Nov/24 07:41
• Updated: 11/Dec/24 09:13
• Resolved: 11/Dec/24 09:13

Description

We currently invalidate the prepared statement cache before making the schema change “live", creating a small window where we might re-prepare a statement with the old schema. This causes us to use the wrong epoch when coordinating a request, making replicas think the coordinator is behind and failing the request until the node is bounced (or prepared statement cache invalidated)

Issue Links

• links to: GitHub Pull Request #3713