[fix][test] Add Delta Tolerance in Double-Precision Assertions to Fix Rounding Flakiness

[LucasEby]

Fixes apache#24970

Motivation

Several tests in this suite compare double values directly using equality assertions. These assertions occasionally fail due to minor inaccuracies introduced by floating-point rounding and double-precision limitations in Java’s IEEE 754 representation. Such discrepancies are expected when performing arithmetic operations on floating-point numbers and can lead to flaky test behavior.

Modifications

A small tolerance (delta) of 1e-5 was added to each double comparison. This delta is at least two orders of magnitude smaller than the smallest compared values, ensuring that it does not affect the logical correctness of the tests. The same delta is already used consistently in another method within the same test class. The tests remain the same, we are just being more careful about double precision issues when comparing values to avoid preventable flaky failures.

Verifying this change

• Make sure that the change passes the CI checks.

This change is already covered by existing tests, such as

• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testNoOwnerLoadData
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testEmptyTopBundlesLoadData
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testRecentlyUnloadedBrokers
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testRecentlyUnloadedBundles
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testSheddingExcludedNamespaces
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testBundlesWithIsolationPolicies
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testBundlesWithAntiAffinityGroup
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testTargetStd
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testSingleTopBundlesLoadData
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testBundleThroughputLargerThanOffloadThreshold
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testZeroBundleThroughput
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testTargetStdAfterTransfer
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testMinBrokerWithLowTraffic
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testMinBrokerWithLowerLoadThanAvg
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testMaxNumberOfTransfersPerShedderCycle
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testLoadBalancerSheddingConditionHitCountThreshold
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testRemainingTopBundles
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testLoadMoreThan100
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testHighVarianceLoadStats
• org.apache.pulsar.broker.loadbalance.extensions.scheduler.TransferShedderTest#testLowVarianceLoadStats

Does this pull request potentially affect one of the following parts:

If the box was checked, please highlight the changes

• Dependencies (add or upgrade a dependency)
• The public API
• The schema
• The default values of configurations
• The threading model
• The binary protocol
• The REST endpoints
• The admin CLI options
• The metrics
• Anything that affects deployment

Documentation

• doc
• doc-required
• doc-not-needed
• doc-complete

Matching PR in forked repository

PR in forked repository:

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!