Es接入Prometheus

环境准备

elasticsearch exporter选用git上开源项目https://github.com/justwatchcom/elasticsearch_exporter

以beta环境为例，beta环境的Es集群信息如下：

	ip	server port
Es node1	10.27.22.153	http://elk9006.eniot.io:9200/
Es node2	10.27.22.154	http://elk9007.eniot.io:9200/
Es node3	10.27.22.155	http://elk9008.eniot.io:9200/

选择最新的版本，本文选用的版本为elasticsearch_exporter-1.1.0.linux-amd64.tar.gz

 

操作流程

1、登录Es node，下载elasticsearch exporter并解压

2、新增/etc/systemd/system/elasticsearch_exporter.service，内容如下（注意开启--es.all参数后会上报其他节点的信息，所以exporter只部署在一台机器上即可）

[Unit] Description=Prometheus elasticsearch_exporter After=local-fs.target network-online.target network.target Wants=local-fs.target network-online.target network.target   [Service] User=root Nice=10 ExecStart=/home/envuser/elasticsearch_exporter-1.1.0/elasticsearch_exporter --es.uri=http://elk9006.eniot.io:9200 --web.listen-address=:9104 --es.timeout=20s --es.all --es.indices_settings --es.cluster_settings ExecStop=/usr/bin/killall elasticsearch_exporter   [Install] WantedBy=default.target

可选参数的作用

usage: elasticsearch_exporter [<flags>]   Flags:   -h, --help                 Show context-sensitive help (also try --help-long and --help-man).       --web.listen-address=":9114"                              Address to listen on for web interface and telemetry.       --web.telemetry-path="/metrics"                              Path under which to expose metrics.       --es.uri="http://localhost:9200"                              HTTP API address of an Elasticsearch node.       --es.timeout=5s        Timeout for trying to get stats from Elasticsearch.       --es.all               Export stats for all nodes in the cluster. If used, this flag will override the flag es.node.       --es.node="_local"     Node's name of which metrics should be exposed.       --es.indices           Export stats for indices in the cluster.       --es.indices_settings  Export stats for settings of all indices of the cluster.       --es.cluster_settings  Export stats for cluster settings.       --es.shards            Export stats for shards in the cluster (implies --es.indices).       --es.snapshots         Export stats for the cluster snapshots.       --es.clusterinfo.interval=5m                              Cluster info update interval for the cluster label       --es.ca=""             Path to PEM file that contains trusted Certificate Authorities for the Elasticsearch connection.       --es.client-private-key=""                              Path to PEM file that contains the private key for client auth when connecting to Elasticsearch.       --es.client-cert=""    Path to PEM file that contains the corresponding cert for the private key to connect to Elasticsearch.       --es.ssl-skip-verify   Skip SSL verification when connecting to Elasticsearch.       --log.level="info"     Sets the loglevel. Valid levels are debug, info, warn, error       --log.format="logfmt"  Sets the log format. Valid formats are json and logfmt       --log.output="stdout"  Sets the log output. Valid outputs are stdout and stderr       --version              Show application version.

 

3、执行命令启动elasticsearch exporter

# systemctl daemon-reload # systemctl start elasticsearch_exporter

4、启动后，exporter会使用指定的metrics端口9104，可以执行如下命令进行验证

# systemctl status -l elasticsearch_exporter ● elasticsearch_exporter.service - Prometheus elasticsearch_exporter    Loaded: loaded (/etc/systemd/system/elasticsearch_exporter.service; disabled; vendor preset: disabled)    Active: active (running) since Thu 2020-01-09 12:13:37 UTC; 2s ago  Main PID: 24699 (elasticsearch_e)    CGroup: /system.slice/elasticsearch_exporter.service            └─24699 /home/envuser/elasticsearch_exporter-1.1.0/elasticsearch_exporter --es.uri=http://elk9006.eniot.io:9200 --web.listen-address=:9104   Jan 09 12:13:37 elk9006 systemd[1]: Started Prometheus elasticsearch_exporter. Jan 09 12:13:37 elk9006 systemd[1]: Starting Prometheus elasticsearch_exporter... Jan 09 12:13:37 elk9006 elasticsearch_exporter[24699]: level=info ts=2020-01-09T12:13:37.530309352Z caller=clusterinfo.go:200 msg="triggering initial cluster info call" Jan 09 12:13:37 elk9006 elasticsearch_exporter[24699]: level=info ts=2020-01-09T12:13:37.530498253Z caller=clusterinfo.go:169 msg="providing consumers with updated cluster info label" Jan 09 12:13:37 elk9006 elasticsearch_exporter[24699]: level=info ts=2020-01-09T12:13:37.533559261Z caller=main.go:148 msg="started cluster info retriever" interval=5m0s Jan 09 12:13:37 elk9006 elasticsearch_exporter[24699]: level=info ts=2020-01-09T12:13:37.533727361Z caller=main.go:188 msg="starting elasticsearch_exporter" addr=:9104

5、修改Prometheus server 上的prometheus.yaml，加入如下内容

# Es_exporter - job_name: 'Es_exporter'   static_configs:   - targets: ['elk9006.eniot.io:9114']

6、在Grafana上新增dashboard，id：2322

效果展示