使用docker和自定义指标搭建promethus服务、alertmanager服务，试用邮件告警功能
$ mkdir -p /Users/xuweiqiang/Documents/alertmanager/
$ mkdir -p /Users/xuweiqiang/Documents/alertmanager/template
$ touch /Users/xuweiqiang/Documents/alertmanager/config.yml
$ touch /Users/xuweiqiang/Documents/alertmanager/template/email.tmpl
$ cd /Users/xuweiqiang/Documents/alertmanagerpackage tool
import (
"github.com/prometheus/client_golang/prometheus"
)
const Application = "application"
const Command = "command"
const Database = "database"
const RequestID = "request_id"
const Status = "status"
type MongoStatusEnum string
const (
Success MongoStatusEnum = "success"
Fail MongoStatusEnum = "fail"
Start MongoStatusEnum = "start"
)
var (
mongoHistogram = prometheus.NewHistogramVec(prometheus.HistogramOpts{
Name: "mongodb_operations_duration_seconds",
Help: "mongodb command record",
Buckets: []float64{0.1, 0.5, 1.0, 1.5, 3.0, 5.0, 10.0},
}, []string{Application, Command, Database, RequestID, Status})
)
func init() {
prometheus.MustRegister(mongoHistogram)
}
func AddMetrics(labels prometheus.Labels, speed float64) {
mongoHistogram.With(labels).Observe(speed)
}public class MybatisMetrics {
private static final String[] labelNames = new String[] { "class", "command", "status" };
public static final Counter QUERY_COUNT = Counter.build()
.name("mybatis_requests_total").help("total sql command.")
.labelNames(labelNames).register();
public static final Gauge QUERY_MAX = Gauge.build()
.name("mybatis_requests_max").help("run sql command latency in seconds.")
.labelNames(labelNames).register();
public static final Summary QUERY_SUMMARY = Summary.build()
.name("mybatis_latency_seconds").help("Request latency in seconds.").labelNames(labelNames)
.register();
}使用docker搭建prometheus和node exporter，解释本机cpu、内存等指标查看方式，以及如何通过http api接口查看监控数据
# 常驻进程启动
$ nohup ./node_exporter > ./node_exporter.log 2>&1 &
# 查看采集到的指标
$ curl 127.0.0.1:9100/metrics$ /Users/xuweiqiang/Documents/dataHCI超融合的监控，我们的应用程序将会和主机一起打包销售，给到客户的时候启动主机，这个主机有可能是1台也有可能是3台。那么主机启动之后，监控服务随之启动,业务会暴露指标给prometheus采集,这里的prometheus如果有1台主机那么就只有1个监控服务，如果有3台主机那么就会有3个监控服务，这3台机器上面的3个prometheus的实例各自的存储是独立的。