[CASSANDRA-16274] Improve performance when calculating StreamTasks with optimised streaming

• Type: Improvement
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 4.0-beta4, 4.0
• Component/s: Consistency/Repair
• Change Category: Performance
• Complexity: Normal
• Platform: All

Source Control Link

https://github.com/apache/cassandra/commit/22abff779df097e0ef38180442e9c680b3d41187

Test and Documentation Plan

new unit tests, jvm dtests

• Assignee: Marcus Eriksson
• Reporter: Marcus Eriksson

Authors

Marcus Eriksson

Reviewers

Sam Tunnicliffe

• Created: 13/Nov/20 10:57
• Updated: 22/Mar/22 21:35
• Resolved: 01/Dec/20 11:42

Description

The way stream tasks are calculated currently is quite inefficient, improve that.

Also, we currently try to distribute the streaming nodes evenly, this creates many more sstables than necessary - instead we should try to stream everything from a single peer, this should reduce the number of sstables created on the out-of-sync node.

Issue Links

• is duplicated by: CASSANDRA-15957 org.apache.cassandra.repair.RepairJobTest testOptimizedCreateStandardSyncTasks
• is related to: CASSANDRA-3200 Repair: compare all trees together (for a given range/cf) instead of by pair in isolation