Skip to content

Conversation

kabo87777
Copy link

Fix Non-Deterministic Behavior in AggregationDistributionTest

Problem

The test testAggregationWithMultiGroupByLevelNode was failing non-deterministically under NonDex with a 40% failure rate due to order-dependent fragment access.

Way to Reproduce

cd iotdb-core/datanode
mvn edu.illinois:nondex-maven-plugin:2.1.1:nondex \
  -Dtest=AggregationDistributionTest#testAggregationWithMultiGroupByLevelNode \
  -DnondexRuns=5

# Expected: Test fails with certain NonDex seeds (e.g., 1016066, 1098954)
# Failure: AssertionError at line 367 when assertions expect specific fragment order

Root Cause

The test accessed fragment instances by fixed array indices (get(0), get(1)), assuming they would always appear in a specific order:

// Assumed fragment 0 has descriptor pattern 1
verifyGroupByLevelDescriptor(expectedDescriptorValue,
    fragmentInstances.get(0)...);

// Assumed fragment 1 has descriptor pattern 2  
verifyGroupByLevelDescriptor(expectedDescriptorValue2,
    fragmentInstances.get(1)...);

When NonDex shuffled collection iteration order during distributed query planning, fragment instances appeared in different orders, causing the test to fail even though the query plan was semantically correct.

Solution

Made the test order-independent by iterating through all fragments and pattern-matching against expected descriptors instead of accessing by index:

  1. Iterate through all fragments instead of assuming fixed indices
  2. Pattern-match descriptors to identify which fragment has which expected pattern
  3. Verify both patterns exist regardless of order
boolean foundFirst = false;
boolean foundSecond = false;

for (FragmentInstance instance : fragmentInstances) {
  // Extract GroupByLevelNode from fragment
  PlanNode childNode = planNodeTree.getChildren().get(0);
  if (!(childNode instanceof GroupByLevelNode)) continue;
  
  GroupByLevelNode groupByLevel = (GroupByLevelNode) childNode;
  if (matchesExpectedDescriptor(groupByLevel, expectedDescriptorValue1)) {
    foundFirst = true;
  } else if (matchesExpectedDescriptor(groupByLevel, expectedDescriptorValue2)) {
    foundSecond = true;
  }
}

assertTrue("Expected to find fragment with single grouped path descriptor", foundFirst);
assertTrue("Expected to find fragment with two specific paths descriptor", foundSecond);

Added helper method matchesExpectedDescriptor() to check if a GroupByLevelNode matches expected descriptor patterns using set-based comparison.


This PR has:

  • been self-reviewed
  • been tested with NonDex to verify non-determinism is eliminated
  • passed all existing tests
  • followed code style guidelines (spotless:apply)

This PR has:

  • been self-reviewed.
    • concurrent read
    • concurrent write
    • concurrent read and write
  • added documentation for new or modified features or behaviors.
  • added Javadocs for most classes and all non-trivial methods.
  • added or updated version, license, or notice information
  • added comments explaining the "why" and the intent of the code wherever would not be obvious
    for an unfamiliar reader.
  • added unit tests or modified existing tests to cover new code paths, ensuring the threshold
    for code coverage.
  • added integration tests.
  • been tested in a test IoTDB cluster.

Key changed/added classes (or packages if there are too many classes) in this PR

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant