[CASSANDRA-12922] Bloom filter miss counts are not measured correctly

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.0.25, 3.11.11, 4.0.1, 4.1-alpha1, 4.1
• Component/s: Legacy/Local Write-Read Paths
• Labels: lhf
• Severity: Normal
• Complexity: Low Hanging Fruit
• Since Version: NA

Source Control Link

https://github.com/apache/cassandra/commit/87424dabd20acfdbb9a4489f289ec075ccfb87b0

Test and Documentation Plan

I added a unit test that covers the changes. No documentation updates are required because it is just a bug fix.

• Assignee: Aleksei Zotov
• Reporter: Branimir Lambov

Authors

Aleksei Zotov

Reviewers

Benjamin Lerer, Branimir Lambov

• Created: 17/Nov/16 09:23
• Updated: 27/May/22 19:24
• Resolved: 19/Jul/21 12:38

Description

Bloom filter hits and misses are evaluated incorrectly in BigTableReader.getPosition: we properly record hits, but not misses. In particular, if we don't find a match for a key in the index, which is where almost all non-matches will be rejected, we don't record a bloom filter false positive.

This leads to very misleading output from e.g. nodetool tablestats.