What problem does it solve?
Manually troubleshooting TiKV performance issues such as write/read latency spikes, throughput drops, and jitter on TiDB Cloud clusters is time-consuming, error-prone, and often leads to misdiagnosis of root causes across TiKV's complex internal layers.
Core Features & Use Cases
- Deterministic Decision Tree: A machine-executable diagnostic flowchart that follows TiKV's layered architecture (from gRPC server down to disk) to systematically identify root causes, eliminating guesswork.
- Version-Aware Metric Catalog: Includes exact PromQL queries for all relevant O11Y platform metrics, with automatic adaptation for TiKV 5.x/6.x vs 7.x+ metric name differences (e.g., raft-engine vs RaftDB metrics).
- Use Case: For example, if your TiDB Cloud cluster experiences sudden write latency spikes, use this skill to run the decision tree to pinpoint if the root cause is disk fsync slowness, write stall, latch contention, PD scheduling disruption, or other layer-specific issues.
Quick Start
Use the tikv-fast-tune skill to diagnose the root cause of TiKV write or read performance issues on your TiDB Cloud cluster by following the built-in layer-aware decision tree.
