pingcap/tidb

Improve observability of fast reorg ADD INDEX progress

Open

#56,986 创建于 2024年10月30日

在 GitHub 查看
 (6 评论) (0 反应) (1 负责人)Go (40,090 star) (6,186 fork)batch import
component/ddlgood first issuetype/enhancement

描述

Enhancement

currently we only let progress forward when KV is written to TiKV. When fast reorg is enabled, data will be batched on disk for a long time. No progress will be made during that stage and some users may suspect DDL is stuck.

So I propose we draw another line in metrics, which means "how many record KV we have read to assume the index KV".

贡献者指南

技术栈
goprometheusgrafana
领域
observabilitydatabase
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
half day
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
active
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
Go programmingTiDB DDLPrometheusKV store
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
50
研究方向
Examine the fast reorg implementation in the DDL package, specifically how index data is batched. Identify the existing metrics (e.g., `ddl worker operation duration`) and determine where to add a new metric for record KV reads. Review the Prometheus client library usage in the codebase. Then implement the new metric and update the Grafana dashboard if applicable.