[CASSANDRA-17272] LeveledCompactionStrategy disk space check improvements

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.0.27, 3.11.13, 4.0.4, 4.1-alpha1, 4.1
• Component/s: Local/Compaction/LCS
• Bug Category: Degradation - Resource Management
• Severity: Normal
• Complexity: Low Hanging Fruit
• Discovered By: Adhoc Test
• Platform: All
• Since Version: 3.0.0

Source Control Link

https://github.com/apache/cassandra/commit/b58a5c86e89e10ad4d39756c5314a756eb18204d

Test and Documentation Plan

cci runs + new unit tests

• Assignee: Marcus Eriksson
• Reporter: Marcus Eriksson

Authors

Marcus Eriksson

Reviewers

Caleb Rackliffe

• Created: 20/Jan/22 14:06
• Updated: 14/Jun/24 00:30
• Resolved: 17/Feb/22 09:49

Description

We currently allow reducing scope (removing sstables from the compaction) when starting STCS-in-L0 with LCS if the compaction is too large for the available disk space. We can do the same for L0 -> L1 compactions - but we can only remove L0 sstables to avoid causing overlap in L1.

Also, in 3.0, when starting an LCS compaction we try to get a writeable location where the full result of the compaction will fit - here we should only get a directory where the first sstable fits, otherwise the compaction might fail even though there is enough space in total among the data directories

Issue Links

• is duplicated by: CASSANDRA-17899 Allow reducing scope for L0 to L1 compactions when space is limited