Skip to content

[Improvement]: Use actualPartitionPlans to determine the optimizingType #3269

@7hong

Description

@7hong

Search before asking

  • I have searched in the issues and found no similar issues.

What would you like to be improved?

In org.apache.amoro.server.optimizing.plan.OptimizingPlanner#planTasks method, using actualPartitionPlans replace evaluators, The resulting optimizingType is more accurate.

for (PartitionEvaluator evaluator : evaluators) {
actualPartitionPlans.add((AbstractPartitionPlan) evaluator);
actualInputSize += evaluator.getCost();
if (actualInputSize > maxInputSize) {
break;
}
}
double avgThreadCost = actualInputSize / availableCore;
List<RewriteStageTask> tasks = Lists.newArrayList();
for (AbstractPartitionPlan partitionPlan : actualPartitionPlans) {
tasks.addAll(partitionPlan.splitTasks((int) (actualInputSize / avgThreadCost)));
}
if (!tasks.isEmpty()) {
if (evaluators.stream()
.anyMatch(evaluator -> evaluator.getOptimizingType() == OptimizingType.FULL)) {
optimizingType = OptimizingType.FULL;
} else if (evaluators.stream()
.anyMatch(evaluator -> evaluator.getOptimizingType() == OptimizingType.MAJOR)) {
optimizingType = OptimizingType.MAJOR;
} else {
optimizingType = OptimizingType.MINOR;
}
}

How should we improve?

No response

Are you willing to submit PR?

  • Yes I am willing to submit a PR!

Subtasks

No response

Code of Conduct

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions