altinity-expert-clickhouse-metrics

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Real-Time Metrics Monitoring

实时指标监控

Real-time monitoring of ClickHouse metrics, events, and asynchronous metrics.

实时监控ClickHouse指标、事件及异步指标。

Diagnostics

诊断

Run all queries from the file checks.sql and analyze the results.

运行checks.sql文件中的所有查询并分析结果。

Ad-Hoc Query Guidelines

临时查询指南

Key Tables

核心表

```
system.metrics
```
- Current gauge values
```
system.events
```
- Cumulative counters since restart
```
system.asynchronous_metrics
```
- System-level metrics
```
system.metric_log
```
- Historical metrics
```
system.asynchronous_metric_log
```
- Historical async metrics

```
system.metrics
```
- 当前仪表盘数值
```
system.events
```
- 重启后的累计计数器
```
system.asynchronous_metrics
```
- 系统级指标
```
system.metric_log
```
- 历史指标
```
system.asynchronous_metric_log
```
- 历史异步指标

Useful Patterns

实用查询模式

sql

-- Find metrics by pattern
select * from system.metrics where metric like '%pattern%'
select * from system.asynchronous_metrics where metric like '%pattern%'
select * from system.events where event like '%pattern%'

sql

-- 按模式查找指标
select * from system.metrics where metric like '%pattern%'
select * from system.asynchronous_metrics where metric like '%pattern%'
select * from system.events where event like '%pattern%'

Cross-Module Triggers

跨模块触发器

Finding	Load Module	Reason
High memory metrics	`altinity-expert-clickhouse-memory`	Memory analysis
High replica delay	`altinity-expert-clickhouse-replication`	Replication issues
High parts count	`altinity-expert-clickhouse-merges`	Merge backlog
High load average	`altinity-expert-clickhouse-reporting`	Query analysis
High connections	`altinity-expert-clickhouse-reporting`	Connection analysis

发现项	加载模块	原因
内存指标过高	`altinity-expert-clickhouse-memory`	内存分析
副本延迟过高	`altinity-expert-clickhouse-replication`	复制问题
分区数量过高	`altinity-expert-clickhouse-merges`	合并积压
负载均值过高	`altinity-expert-clickhouse-reporting`	查询分析
连接数过高	`altinity-expert-clickhouse-reporting`	连接分析

Monitoring Recommendations

监控建议

Key Metrics to Alert On

需设置告警的核心指标

Metric	Warning	Critical
`ReadonlyReplica`	-	> 0
`Query`	> 75% max	> 90% max
`MemoryResident`	> 80% RAM	> 90% RAM
`MaxPartCountForPartition`	> parts_to_delay	> parts_to_throw
`ReplicasMaxAbsoluteDelay`	> 5 min	> 1 hour
`LoadAverage1`	> CPU count	> 2x CPU count

指标	警告阈值	严重阈值
`ReadonlyReplica`	-	> 0
`Query`	> 最大阈值的75%	> 最大阈值的90%
`MemoryResident`	> 内存的80%	> 内存的90%
`MaxPartCountForPartition`	> parts_to_delay	> parts_to_throw
`ReplicasMaxAbsoluteDelay`	> 5分钟	> 1小时
`LoadAverage1`	> CPU核心数	> 2倍CPU核心数

Prometheus/Grafana Export

Prometheus/Grafana 导出

ClickHouse exposes metrics at

:9363/metrics

in Prometheus format when enabled.

当启用后，ClickHouse会在

:9363/metrics

地址以Prometheus格式暴露指标。