如何解决Prometheus chart Ingres/Service映射？

Question

刚刚安装了具有以下值的 stable/prometheus 图表，我可以从 pods 但不能从主机的网络浏览器访问服务器前端。

我的values.yaml:

alertmanager:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - localhost/alerts

server:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - localhost/prom

pushgateway:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - localhost/push

我使用 nginx ingress 并创建了 ingress，但由于某些未知原因，它似乎没有映射到服务。

一些数据：

我可以通过默认和 dns 服务名称从入口 pods（以及所有其他）访问服务器：

kubectl exec -it nginx-ingress-controller-5cb489cd48-t4dgv -- sh
/etc/nginx $ curl prometheus-server.default.svc.cluster.local
<a href="/graph">Found</a>

/etc/nginx $ curl prometheus-server
<a href="/graph">Found</a>

图表创建的活动入口列表：

kubectl get ingress
NAME                      HOSTS       ADDRESS     PORTS   AGE
nginx-ingress             localhost   localhost   80      37h
prometheus-alertmanager   localhost   localhost   80      43m
prometheus-pushgateway    localhost   localhost   80      43m
prometheus-server         localhost   localhost   80      43m

活动服务资源列表：

kubectl get svc
NAME                            TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)                      AGE
kubernetes                      ClusterIP      10.96.0.1        <none>        443/TCP                      37h
nginx-deployment                ClusterIP      10.100.1.167     <none>        80/TCP                       37h
nginx-ingress-controller        LoadBalancer   10.109.57.131    localhost     80:32382/TCP,443:30669/TCP   36h
nginx-ingress-default-backend   ClusterIP      10.107.91.35     <none>        80/TCP                       36h
php-deployment                  ClusterIP      10.105.73.26     <none>        9000/TCP                     37h
prometheus-alertmanager         ClusterIP      10.97.89.149     <none>        80/TCP                       44m
prometheus-kube-state-metrics   ClusterIP      None             <none>        80/TCP,81/TCP                44m
prometheus-node-exporter        ClusterIP      None             <none>        9100/TCP                     44m
prometheus-pushgateway          ClusterIP      10.105.81.111    <none>        9091/TCP                     44m
prometheus-server               ClusterIP      10.108.225.187   <none>        80/TCP                       44m

另一方面，如果我将子域声明为入口主机，则可以访问 Prometheus：

alertmanager:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - alerts.localhost

server:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - prom.localhost

pushgateway:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - push.localhost

是我做错了什么还是这有什么问题？有什么建议吗？

提前致谢！

Helm 和 Kubernetes 版本： Helm 3.0.3 / Kubernetes 1.15.5（Docker for Mac，MacOS Catalina）

Answer 1

我重现了您的场景，并且通过运行一些测试，我了解到这不会按照您希望的方式工作。这不是实施它的正确方法。

让我们深入研究一下。

您可以在入口中添加 nginx.ingress.kubernetes.io/rewrite-target，如本例所示：

$ kubectl get ingresses. myprom-prometheus-pushgateway -oyaml
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  annotations:
    kubernetes.io/ingress.class: nginx
    nginx.ingress.kubernetes.io/rewrite-target: /
  creationTimestamp: "2020-02-18T09:51:32Z"
  generation: 1
  labels:
    app: prometheus
    chart: prometheus-10.4.0
    component: pushgateway
    heritage: Helm
    release: myprom
  name: myprom-prometheus-pushgateway
  namespace: default
  resourceVersion: "3239"
  selfLink: /apis/extensions/v1beta1/namespaces/default/ingresses/myprom-prometheus-pushgateway
  uid: 499372f4-52b1-4b37-982c-b52e70657d37
spec:
  rules:
  - host: localhost
    http:
      paths:
      - backend:
          serviceName: myprom-prometheus-pushgateway
          servicePort: 9091
        path: /push
status:
  loadBalancer:
    ingress:
    - ip: 192.168.39.251

添加后，您将能够按预期访问。不幸的是，这样做之后你将面临一个新问题。如果我们检查 curl 命令的 html 输出，我们可以看到：

<!DOCTYPE html>

<html lang="en">
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
    <meta name="robots" content="noindex,nofollow">
    <title>Prometheus Pushgateway</title>

    <link rel="shortcut icon" href="/static/favicon.ico?v=793293bdadd51fdaca69de5bb25637b0f93b656b">

    <script src="/static/jquery-3.4.1.min.js?v=793293bdadd51fdaca69de5bb25637b0f93b656b"></script>
    <script src="/static/bootstrap-4.3.1-dist/js/bootstrap.min.js?v=793293bdadd51fdaca69de5bb25637b0f93b656b"></script>
    <script src="/static/functions.js?v=793293bdadd51fdaca69de5bb25637b0f93b656b"></script>

    <link type="text/css" rel="stylesheet" href="/static/bootstrap-4.3.1-dist/css/bootstrap.min.css?v=793293bdadd51fdaca69de5bb25637b0f93b656b">
    <link type="text/css" rel="stylesheet" href="/static/prometheus.css?v=793293bdadd51fdaca69de5bb25637b0f93b656b">
    <link type="text/css" rel="stylesheet" href="/static/bootstrap4-glyphicons/css/bootstrap-glyphicons.min.css?v=793293bdadd51fdaca69de5bb25637b0f93b656b">
  </head>

可以看出，我们有 /static/jquery-3.4.1.min.js 示例，所以这会将您的浏览器重定向到不同的位置。问题是你没有这个位置。

这就是为什么我建议您避免使用路径并坚持使用涉及使用子域的辅助解决方案。

alertmanager:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - alerts.localhost

server:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - prom.localhost

pushgateway:
  ingress:
    enabled: true
    annotations:
      kubernetes.io/ingress.class: nginx
    hosts:
      - push.localhost

如何解决Prometheus chart Ingres/Service映射？

How to solve Prometheus chart Ingres/Service mapping?

kubernetes

prometheus

kubernetes-helm

nginx-ingress