[CASSANDRA-15058] Avoid double counting read latencies for digest queries

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.0.19, 3.11.5, 4.0-alpha1, 4.0
• Component/s: Observability/Metrics
• Bug Category: Correctness - API / Semantic Implementation
• Severity: Normal
• Complexity: Low Hanging Fruit
• Discovered By: Workload Replay
• Since Version: 3.0 alpha 1

Test and Documentation Plan

A distributed in-jvm test was added to make sure we don't overcount

• Assignee: Marcus Eriksson
• Reporter: Marcus Eriksson

Authors

Marcus Eriksson

Reviewers

Sam Tunnicliffe

• Created: 19/Mar/19 15:45
• Updated: 15/May/20 07:59
• Resolved: 26/Mar/19 12:26

Description

We are closing the UnfilteredPartitionIterator wrapped with withMetricsRecording twice when we get digest requests - closing it calls onClose and that makes the metrics update twice.