一、背景介绍
如果我们的物理机有很多，不管是基于"file_sd_config"还是"kubernetes_sd_config"，我们都需要手动写入目标target及创建目标service，这样才能被prometheus自动发现，为了避免重复性工作过多，所以我们可以使用Consul做注册中心，让所有的机器注册至Consul，随后Prometheus再从Consul上来发现目标主机完成自动加入Prometheus的Target中。
二、Consul部署
Consul部署至K8s集群内部，所以直接使用helm部署即可。
- 获取consul的values
# 增加helm的repo
~] helm repo add hashicorp https://helm.releases.hashicorp.com
# 查看repo是否添加成功
~] helm repo list
NAME URL
hashicorp https://helm.releases.hashicorp.com
# 获取consul的values,具体参数含义可参考官网
# https://developer.hashicorp.com/consul/docs/k8s/helm
~] helm inspect values hashicorp/consul &> /gensee/k8s_system/consul/
~] vim /gensee/k8s_system/consul/values.yaml
server:
...
# 这里是定义部署3个consul以实现consul集群，保证数据高可用
replicas: 3
# 要与server.replicas的值一致
bootstrapExpect: 3
# 定义每个节点使用多少的存储空间存储数据,默认100Gi
storage: 100Gi
# 定义storageclass的名称，如果有的话
storageClass: managed-nfs-storage
...- 创建名称空间并安装consul
~] kubectl create ns consul
~] helm install consul hashicorp/consul --namespace consul --values /gensee/k8s_system/consul/values.yaml
# 查看consul安装状态
~] helm status consul -n consul
~] kubectl get all -n consul
NAME READY STATUS RESTARTS AGE
pod/consul-consul-connect-injector-89d7cf9d8-vqzgm 1/1 Running 7 3d23h
pod/consul-consul-server-0 1/1 Running 0 3d22h
pod/consul-consul-server-1 1/1 Running 0 3d22h
pod/consul-consul-server-2 1/1 Running 0 3d22h
pod/consul-consul-webhook-cert-manager-649d7486d7-x2pn5 1/1 Running 0 3d23h
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/consul-consul-connect-injector ClusterIP 172.13.185.211 <none> 443/TCP 3d23h
service/consul-consul-dns ClusterIP 172.13.152.148 <none> 53/TCP,53/UDP 3d23h
service/consul-consul-server ClusterIP None <none> 8500/TCP,8502/TCP,8301/TCP,8301/UDP,8302/TCP,8302/UDP,8300/TCP,8600/TCP,8600/UDP 3d23h
# 对外提供UI的svc
service/consul-consul-ui ClusterIP 172.4.246.29 <none> 80/TCP 3d23h
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/consul-consul-connect-injector 1/1 1 1 3d23h
deployment.apps/consul-consul-webhook-cert-manager 1/1 1 1 3d23h
NAME DESIRED CURRENT READY AGE
replicaset.apps/consul-consul-connect-injector-89d7cf9d8 1 1 1 3d23h
replicaset.apps/consul-consul-webhook-cert-manager-649d7486d7 1 1 1 3d23h
NAME READY AGE
statefulset.apps/consul-consul-server 3/3 3d23h三、Nginx代理Consul
由于Consul部署在K8s集群内部，且集群并没有直接对外，所以使用nginx代理到集群内部Consul以提供UI界面
server {
listen 80;
server_name xxx-xxx.xxxxx.xxx;
return 301 https://$host$request_uri;
}
server {
# For https
listen 443 ssl;
# listen [::]:443 ssl ipv6only=on;
ssl_certificate sslkey/server.cer;
ssl_certificate_key sslkey/server.key;
server_name xxx-xxx.xxxxx.xxx; # 替换成你的域名
access_log logs/consul-access.log;
error_log logs/consul-error.log;
location / {
# 这个svc是consul-ui
proxy_pass http://consul-consul-ui.consul.svc.cluster.local;
# 增加nginx自带basic认证
auth_basic "Basic Authentication";
auth_basic_user_file "/etc/nginx/conf/system/.htpasswd";
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header Host $host;
proxy_set_header X-Forward-For $proxy_add_x_forwarded_for;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection 'upgrade';
}
}四、注册至Consul
# id和tags的第二个字段是主机名,xxx.xx.x.xxx是主机的ip地址,其他保留
~] curl -u xxxxx:xxxxx -X PUT -d \
'{"id": "xxx-xxx-xxxx","name": "node_exporter","address": "xxx.xx.x.xxx","port": 9100,"tags":["prometheus","xxx-xxx-xxxx"],"checks": [{"http": "http://xxx.xx.x.xxx:9100/metrics", "interval": "5s"}]}' \
https://xxx-xxx.xxxxx.xxx/v1/agent/service/register
# 从consul中注销,node-exporter是id，如果注销失败，说明第一次请求没有到这台机器所在的consul节点，需要再次执行
~] curl -X PUT http://xxx-xxx.xxxxx.xxx/v1/agent/service/deregister/node-exporter五、prometheus配置consul_sd_config
# 与此文档(https://www.yuque.com/kkkfree/itbe1d/zzzgu0)中的第四.1一样，修改prometheus-additional.yaml，在kubernetes_sd_configs下面添加
- job_name: 'consul-prometheus'
consul_sd_configs:
# 这是可以直接调用consul的svc了，不用写ui了，因为都在K8s集群中
- server: 'consul-consul-server.consul.svc.cluster.local:8500'
services:
# 指定获取哪个service，我这里因为机器都注册到了node_exporter这个service，所以我只抓取这个
- 'node_exporter'
relabel_configs:
# 将注册上来的id标签替换为hostname，看起来比较直观
- source_labels: [__meta_consul_service_id]
action: replace
target_label: hostname
# 重载prometheus配置文件
~] kubectl delete secret generic additional-configs -n monitoring
~] kubectl create secret generic additional-configs --from-file=prometheus-additional.yaml -n monitoring
# 手动使prometheus重新读取配置
~] kubectl get pods -o wide -n monitoring
~] kubectl get pods -o wide -l app=prometheus -n monitoring
~] curl -X POST http://IP:9090/-/reload六、效果图
- consul的node_exporter的service中实例展示
- prometheus自动发现的endpoint
