schema_version: 1
id: latency-surge-with-tikv-cpu-imbalance
title: Latency surge with TiKV CPU imbalance
status: draft
owner: codex
source: jira
environment:
  profile: stable-dedicated
  cluster_id: '10176671098661309318'
  cluster_type: dedicated
  tidb_version: v8.5.4
time_window:
  incident_at: '2026-08-06T06:50:00Z'
  allowed:
  - id: reported_incident_window
    start: '2026-08-06T06:30:00Z'
    end: '2026-08-06T07:30:00Z'
prompt:
  text: 请分析集群 10176671098661309318 在 2026-08-06T06:30:00Z 到 2026-08-06T07:30:00Z 的延迟和 HTTP 500 激增。一个 TiKV 节点 CPU 在 06:54Z 左右接近 100%，某个高 QPS 表的 region leader 明显偏向该节点，TopSQL 中也有大量回表读取。请判断主因、数据分布因素以及处理建议。
tags:
- sql
- latency
- tikv-cpu
case_tier: P1
cost_tier: medium
followup:
  scheduled_max_followups: 0
execution:
  acp_bash_timeout_s: 900