【Prometheus】2. Overview and deployment

summary

1. Prometheus What is it?

brief introduction

Prometheus By SoundCloud Open source monitoring alarm solution .2012 Became open source in the community , Have a very active developer and user community ,Prometheus On 2016 Joined cloud native Computing Foundation in （CNCF）, To become a successor k8s And then the second hosting project .

https://prometheus.io/

https://github.com/prometheus

characteristic

• Multidimensional data model ： Time series data indicated by the measurement name and key value pair
• PromSQL： A flexible query language , Complex queries can be performed using multidimensional data
• Independent of distributed storage , A single service node can work directly
• be based on HTTP Of pull Method to collect time series data
• Push time series data through PushGateWay Component support
• Discover targets through service discovery or static configuration
• Various graphic models and dashboard support

2. Prometheus Composition and architecture

• Data collection ：
  • Push gateway： Short term storage of indicator data , Mainly used for temporary tasks
  • jobs/exporters： Collect the existing monitoring indicators of third-party services and expose them metrics, similar sql/http Mission , adopt jobs( similar http Will expose metrics Interface supply Prometheus Grab ) perhaps exporters( monitor agent, Get different data sources , Re pass Prometheus serve Grab )
• Prometheus Server： Collect and store time series databases , And provide query interface
• Service discovery： Service discovery and monitoring native support k8s All round monitoring
• WEB UI： adopt PromQL View the data ,Prometheus web UI（ ordinary web Console ）,grafana,API clients
• Alertmanager： Monitoring alarm assembly , Support email/wechat
• ClientLibray： Client library , Support python java golang etc. , Support exposure metrics

3. Data model

• Prometheus Store all data as time series , Have the same measurement name and label belong to the same indicator ;
• Each time series has a metric name and a set of key value pairs （ Also known as label ）
• Time series format ： <metric name>{<label name>=<label value>, ...}

Example ：api_http_requests_total{method="POST", handler="/messages"}

4. The index type

• counter： Incremented counter
• gauge： Any number that can be changed
• histogram： Sampling data over a period of time , And sum up all data values and count the quantity
• summary： And histogram similar

5. Assignments and examples

example ： Objects that can be captured become instances （Instances）

Homework ： A collection of instances with the same target is called a job （jobs）

Deploy

1. Binary deployment

• Download binary package ：prometheus.io
• decompression prometheus：tar -zxvf prometheus-2.26.0.linux-amd64.tar.gz
• Modify working directory ：mv prometheus-2.26.0.linux-amd64 /usr/local/prometheus
• Enter working directory ：cd /usr/local/prometheus
• Check the parameter description ：./prometheus --help
• Configure the service :
  • cd /usr/lib/systemd/system
  • vim prometheus.service
  • systemctl daemon-reload

vim prometheus.service

• Start the service ：systemctl start prometheus
• Out of Service ：systemctl stop prometheus

2. docker Deploy

docker run \ -p 9090:9090 \ -v /path/to/prometheus.yml:/etc/prometheus/prometheus.yml \ prom/prometheus