[KAFKA-5428] Transactional producer aborts batches incorrectly in abortable error state

• Type: Sub-task
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 0.11.0.0
• Component/s: clients, core, producer
• Labels: exactly-once
• Assignee: Jason Gustafson
• Reporter: Jason Gustafson
• Created: 11/Jun/17 07:09
• Updated: 13/Jun/17 01:10
• Resolved: 13/Jun/17 01:10

Description

We currently abort batches blindly if we are in any error state. We should only do this if we are in a fatal error state. Otherwise, we risk OutOfOrderSequence errors if a failed produce request had actually been written successfully to the topic.

Issue Links

• links to: GitHub Pull Request #3298