SS

starrocks/starrocks-debug-skills

开发工具

本页由 AI 翻译,技术细节请以原文为准。 查看原文

78 stars 质量 57 趋势 18Apache-2.0

星岩集群诊断与故障排除技能

概览

星岩集群诊断与故障排除技能 星岩集群诊断和故障排除技能系统。 该技能系统化了故障排查经验,帮助为集群问题提供结构化的调查指导,涵盖常见生产问题,并从症状到根本原因的完整调查路径。 每个技能目录包含:- — 结构化故障排除说明(主要入口) - — 针对技能的参考文档和案例研究 - — 技能专用辅助脚本(仅限查询/、节点/、CPU-饱和/) 1. — 客户端→FE→→存储/网络 2.— 有日志、指标、栈跟踪支持3.— 通过参数调优实现服务恢复 4.

概述(AI 翻译)


类型:README 描述:星岩集群诊断与故障排除技能

星岩调试技能

星岩集群诊断和故障排除技能系统。

概述

该技能系统化了故障排查经验,帮助为集群问题提供结构化的调查指导,涵盖常见生产问题,并从症状到根本原因的完整调查路径。

目录结构

每个技能目录包含:

  • SKILL.md — 结构化故障排除指令(主要入口)
  • 参考文献/ — 按技能分类的参考文档和案例研究
  • 脚本/ — 技能专用辅助脚本(仅限查询/、节点/、cpu-saturation/)

技能类别

类别 关键词 描述
查询 挂机、减速、剖面、扫描、加入 查询故障排除,慢查询分析
导入 超时,RPC失败,发布 导入故障排查,代理/流/例程加载
节点 崩溃、OOM、死锁、GC FE/BE节点问题、内存分析
materialized-view 刷新、超时、重写 MV 刷新失败、查询重写
数据湖 HMS,Kerberos,外部 外部目录,蜂巢,冰山,S3

本页由 AI 翻译,技术细节请以原文为准。

README

以下为原文 README 节选,完整内容请查看原文。


type: readme description: StarRocks cluster diagnostics and troubleshooting skill

starrocks-debug-skills

StarRocks cluster diagnostics and troubleshooting skill system.

Overview

This skill systematizes troubleshooting experience to help provide structured investigation guidance for cluster issues, covering common production problems with complete investigation paths from symptom to root cause.

Directory Structure

starrocks-debug-skills/
├── README.md
├── LICENSE
├── CONTRIBUTING.md
├── guides/                     # Cross-skill cascade guides
│   ├── cascade-cluster-degradation.md
│   └── cascade-import-rpc-failed.md
├── references/                 # Shared commands used across all skills
│   └── 01-common-commands.md
├── scripts/                    # Utility scripts
│   ├── install.sh              # Install skills to IDEs
│   ├── skill_converter.py      # Verify structure, convert to IDE formats
│   └── analyze_logs.py         # Audit log analysis script
├── query/                      # Query troubleshooting
│   ├── SKILL.md
│   ├── references/
│   │   ├── 02-profile-and-sql-toolkit.md
│   │   ├── 03-statistics-index.md
│   │   └── case-014-scan-skew.md
│   └── scripts/
│       └── analyze_logs.py
├── import/                     # Import troubleshooting
├── node/                       # BE/FE node issues
├── materialized-view/          # MV troubleshooting
├── data-lake/                  # External catalog issues
├── shared-data/                # Shared-data (lake-mode) architecture
├── tablet/                     # Tablet health issues
├── deployment/                 # FE/BE startup and configuration
├── high-concurrency/           # High QPS optimization
├── resource-isolation/         # Resource groups and query queues
├── balance/                    # Tablet scheduler and cluster balance
├── compaction/                 # Compaction issues
├── cpu-saturation/             # CPU triage and attribution
├── .claude/
├── .cursor/
└── .idea/

Each skill directory contains:

  • SKILL.md — Structured troubleshooting instructions (main entry point)
  • references/ — Per-skill focused reference docs and case studies
  • scripts/ — Skill-specific helper scripts (query/, node/, cpu-saturation/ only)

Skills Categories

Category Keywords Description
query hang, slow, profile, scan, join Query troubleshooting, slow query analysis
import timeout, RPC failed, publish Import troubleshooting, broker/stream/routine load
node crash, OOM, deadlock, GC FE/BE node issues, memory analysis
materialized-view refresh, timeout, rewrite MV refresh failures, query rewrite
data-lake HMS, Kerberos, external External catalog, Hive, Iceberg, S3
shared-data DataCache, S3, leader switch Shared-data (lake-mode) deployments only
tablet health, balance, skew Tablet health, disk balancing
在 GitHub 查看完整 README

推荐工具

换一个关键词,或者移除筛选条件。

安装

npx skillfish add starrocks/starrocks-debug-skills
新手?先看 mcpskill.org 教程 →