Prometheus External Metric not working

[zalando-incubator]

Prometheus External Metric not working

[mikkeloscar]

Hi, could you post the full log output of the kube-metrics-adapter from when you start it till you start to see those errors?

It would also be good if you could post the deployment.yaml of the final configuration. It's hard to determine the actual configuration just by looking at the chart, since there are many configuration variables.

[veetow]

Sure thing, here are the logs:
https://gist.github.com/veetow/ac91b34643f4c1678e990baa0f8f810e

Here is the deployment:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: kube-metrics-adapter
  labels:
    app: kube-metrics-adapter
    chart: kube-metrics-adapter-0.0.4
    release: "kube-metrics-adapter"
    heritage: "Tiller"
spec:
  replicas: 1
  selector:
    matchLabels:
      app: kube-metrics-adapter
      release: "kube-metrics-adapter"
  template:
    metadata:
      labels:
        app: kube-metrics-adapter
        chart: kube-metrics-adapter-0.0.4
        release: "kube-metrics-adapter"
        heritage: "Tiller"
    spec:
      serviceAccountName: kube-metrics-adapter
      securityContext:
        runAsNonRoot: true
        runAsUser: 10001
      containers:
      - name: kube-metrics-adapter
        image: "banzaicloud/kube-metrics-adapter:0.1.1"
        imagePullPolicy: IfNotPresent
        args:
        - --secure-port=8443
        - --v=9
        - --prometheus-server=http://prometheus-prometheus-oper-prometheus.monitoring.svc.cluster.local:9090
        - --cert-dir=/tmp/cert
        ports:
        - containerPort: 8443
          name: https
        securityContext:
          allowPrivilegeEscalation: false
          capabilities:
            drop: ["all"]
          readOnlyRootFilesystem: true
        volumeMounts:
        - mountPath: /tmp
          name: tmp
      volumes:
      - name: tmp
        emptyDir: {}

[mikkeloscar]

Ok, I see, you are using the image: banzaicloud/kube-metrics-adapter:0.1.1 which is from their fork (https://github.com/banzaicloud/kube-metrics-adapter/). This doesn't include the latest features like external prometheus metrics.

You can either use the latest image we have built from master:

registry.opensource.zalan.do/teapot/kube-metrics-adapter:master-35

Or build from source if you prefer: IMAGE=$YOURNAME/kube-metrics-adapter make build.push

[veetow]

Ah, I couldn't find a public image published by zalando. I'll give this one a go.

[veetow]

Using that image seems to have solved my issue. Any plans to push those to quay or docker hub?

[mikkeloscar]

Yeah, sorry about not making this clear. We will start doing releases and include a reference to the published images.

[mikkeloscar]

Will track creating releases in #16

[veetow]

Thanks!