Flink 指标监控整理

本篇文章对Flink的指标监控进行讲解，期望能够帮助到大家更好的对Flink进行监控。

1.整体流程 

整体的流程包括3个部分，指标推送->采集指标->展示指标。我们选择使用PushGateway来推送指标，Prometheus来采集及存储指标，Grafana来配置Dashboard，让指标进行可视化。

2.软件安装及配置

2.1PushGateway

我们在测试环搭建已经搭建好了Promethues及Grafana，详细步骤参考Apache Airflow指标监控实践,再搭建一下PushGateway就行了。

-- 下载pushgateway软件包
wget -c
https://github.com/prometheus/pushgateway/releases/download/v0.9.1/pushgateway-0.9.1.linux-amd64.tar.gz
-- 启动pushgateway
nohup ./pushgateway --web.listen-address :9091 > /var/log/pushgateway.log 2>&1 &

再在prometheus.yml加入pushgateway的配置

# scrape_configs 下面加入刚刚启动的pushgateway信息
- job_name: 'pushgateway'
    static_configs:
    - targets: ['cdh-datanode01:9091']
      labels:
        instance: pushgateway

 重启prometheus，检查配置是否OK

http://prometheus_host:9090/targets

显示Status为Up则说明是OK的

2.2 Flink 1.13

 

3.启动测试Flink程序

# 打开 9999 socket端口
nc -l 9999
# 启动Flink自带的 SocketWindowWordCount程序
/opt/flink-1.13.0/bin/flink run -m yarn-cluster --yarnqueue root.analysis -ynm socketWindowWordCount -p 1 -yjm 1024m -ytm 3072m /opt/flink-1.13.0/examples/streaming/SocketWindowWordCount.jar -d --host cdh-datanode01 --port 9999

4.指标说明

 

5.Grafana Dashboard配置及展示

 

参考文档

[0] 使用Prometheus+Grafana监控Flink on YARN作业

https://www.jianshu.com/p/886378855cea

[1] Flink Metrics

https://ci.apache.org/projects/flink/flink-docs-release-1.13/docs/ops/metrics/

[2] Flink Metric Reporters

https://ci.apache.org/projects/flink/flink-docs-release-1.13/docs/deployment/metric_reporters/

[3] 腾讯云Prometheus - Flink 接入

https://cloud.tencent.com/document/product/248/50974