CNDB-15058: track in-flight BF memory usage from trie index writer if early open is not enabled (#1949)

[djatnieks]

What is the issue

cndb-15058: table metrics - bloom filter memory usage doesn't include partially written sstables' if early open is not enabled

What does this PR fix and why was it fixed

Include in flight BF memory usage if early open is not enabled

[github-actions]

Checklist before you submit for review

• This PR adheres to the Definition of Done
• Make sure there is a PR in the CNDB project updating the Converged Cassandra version
• Use NoSpamLogger for log lines that may appear frequently in the logs
• Verify test results on Butler
• Test coverage for new/modified code is > 80%
• Proper code formatting
• Proper title for each commit staring with the project-issue number, like CNDB-1234
• Each commit has a meaningful description
• Each commit is not very long and contains related changes
• Renames, moves and reformatting are in distinct commits
• All new files should contain the DataStax copyright header instead of the Apache License one

[djatnieks]

@jasonstack The 5.0 implementation of bf metrics is changed and I think this is causing the test to fail - I left a FIXME comment in BloomFilterMetrics which seems to me the closest equivalent of TableMetrics.bloomFilterOffHeapMemoryUsed createTableGuage implements getValue` with:

     long total = inFlightBloomFilterOffHeapMemoryUsed.get();

In 5.0 I cannot see a way to access inFlightBloomFilterOffHeapMemoryUsed from TableMetrics in BloomFilterMetrics.

I considered moving inFlightBloomFilterOffHeapMemoryUsed to BloomFilterMetrics, but TableMetrics is per-table while BloomFilterMetrics is not and currently only operates on sstable readers, so that didn't seem to help any.

I want to check with you about options before I consider something else ... let me know what you think - thanks!

[jasonstack]

Hi @djatnieks the refactoring in 5.0 is tricky..

Maybe we can refactor GaugeProvider to use Function<ColumnFamilyStore, T> neutralValue to extract value from inFlightBloomFilterOffHeapMemoryUsed (which can be stored in CFS or TableMetrics).

Similar to the following, but it uses a fixed neutralValue

    protected final <T extends Number> GaugeProvider<T> newGaugeProvider(String name, T neutralValue, Function<R, T> extractor, BiFunction<T, T, T> combiner)
    {
        return new SimpleGaugeProvider<>(this::map, name, readers -> {
            T total = neutralValue;
            for (R reader : readers)
                total = combiner.apply(total, extractor.apply(reader));
            return total;
        });
    }

[sonarqubecloud]

Quality Gate passed

Issues
2 New issues
0 Accepted issues

Measures
0 Security Hotspots
89.2% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[cassci-bot]

❌ Build ds-cassandra-pr-gate/PR-1967 rejected by Butler

---

9 regressions found
See build details here

---

Found 9 new test failures

Test	Explanation	Runs	Upstream
o.a.c.cql3.CqlBuilderTest.testAppendOptionsChaining (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsSortsMapKeys (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsWithEmptyOptions (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsWithExistingContent (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsWithMultipleOptions (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsWithSingleOption (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testAppendOptionsWithSpecialCharacters (compression)	REGRESSION	🔴	0 / 1
o.a.c.cql3.CqlBuilderTest.testOptionsBuilderWithMixedEmptyAndNonEmpty (compression)	REGRESSION	🔴	0 / 1
o.a.c.index.sai.cql.GeoDistanceAccuracyTest.haversineBenchmark (compression)	REGRESSION	🔴	0 / 1

---

Found 14 known test failures

[djatnieks]

the refactoring in 5.0 is tricky..

Yes, indeed. Thanks for the suggestions @jasonstack - I was able to update AbstractMetricsProviders with a new guage provider to use for bloomFilterOffHeapMemoryUsed that uses inFlightBloomFilterOffHeapMemoryUsed and now the new test is passing.