[CASSANDRA-17834] Repair coordinator can get stuck in an infinite loop if it gets a FailSession after it's marked a session FINALIZED

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 4.0.6, 4.1-beta1, 5.0-alpha1, 5.0
• Component/s: Consistency/Repair
• Bug Category: Degradation - Other Exception
• Severity: Low
• Complexity: Low Hanging Fruit
• Discovered By: User Report
• Platform: All
• Since Version: 4.0

Source Control Link

https://gitbox.apache.org/repos/asf?p=cassandra.git;a=commit;h=0353df7542dbdbb1140a72899666e4587e87a083

Test and Documentation Plan

New distributed testing to illustrate behavior and fix

• Assignee: Josh McKenzie
• Reporter: Josh McKenzie

Authors

Josh McKenzie

Reviewers

David Capwell

• Created: 18/Aug/22 16:52
• Updated: 12/Sep/23 13:03
• Resolved: 22/Aug/22 16:52

Description

If the repair coordinator gets a FailSession message after it has marked the session as FINALIZED it will start an infinite loop of FailSession messages. This is unique to 4.0+ in CoordinatorSession.java

Issue Links

• causes: CASSANDRA-17843 Fix test/distributed/org/apache/cassandra/distributed/test/IncRepairCoordinatorErrorTest