[CASSANDRA-19718] CEP-15: (Accord) SyncPoint timeouts become a Exhausted rather than a Timeout and doesn’t get retried

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: NA
• Component/s: Accord
• Bug Category: Correctness - API / Semantic Implementation
• Severity: Low
• Complexity: Low Hanging Fruit
• Discovered By: Fuzz Test
• Platform: All
• Since Version: NA

Source Control Link

https://github.com/apache/cassandra/commit/0ef232f88b83a8d1bba6e406c9075e9e051d0241

Test and Documentation Plan

new testing

• Assignee: David Capwell
• Reporter: David Capwell

Authors

David Capwell

Reviewers

Ariel Weisberg

• Created: 18/Jun/24 20:30
• Updated: 18/Jun/24 23:30
• Resolved: 18/Jun/24 23:30

Description

In Cassandra we try to make sure coordinators return timeout if every call under it was also a timeout, this makes it easier to understand what is going on (coordination failure due to timeouts looks very different than us just timing out), but accord doesn't do this; leading to an Exhausted error (which we don't retry)

Issue Links

• links to: GH: accord trunk
• links to: GH: cep-15-accord