Skip to content
This repository has been archived by the owner on Apr 12, 2024. It is now read-only.

Latest commit

 

History

History
420 lines (339 loc) · 17.2 KB

getting-started.md

File metadata and controls

420 lines (339 loc) · 17.2 KB
Raw

Getting started

Overview

vSAN Prometheus provides essential container images for monitoring vSAN infrastructure using Prometheus & Grafana.

This page provides examples for deploying vSAN monitoring solutings using Prometheus & Grafana. Our examples may not be ideal for (existing) production environments, but should highlight how our service discovery (via sidecar or Prometheus Operator) can be integrated into any Prometheus environment.

The highlights for deploying our sample vSAN monitoring solutions:

  • A full end to end vSAN monitoring solution based on Prometheus & Grafana.
  • Using Helm charts for deploying our solution.
  • The deployment solution considers vSAN 7.0 and vSAN pre 7.0 versions.
  • Sample Grafana dashboards are provided in the deployment.

Background

In this repo, we highlight Prometheus and Grafana for vSAN metrics monitoring. The background document presents why Prometheus and Grafana are good for vSAN metrics monitoring.

Prerequisites

vSAN Prometheus solution runs on Docker and Kubernetes environment. We test the Docker v18.09.9 and Kubernetes v1.15.4 for running our deployment solution.

We suggest you can use the latest version of Docker and Kubernetes. Since our test cannot cover any newer version of Docker and Kubernetes, please file an issue about it. We also keep updating this project for supporting new release of Docker and Kubernetes.

Helm

Our solution is based on Helm 3. We suggest to use the stable helm repository.

Install Helm 3 using Homebrew in Mac and add stable repository.

brew install helm
helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm repo add grafana https://grafana.github.io/helm-charts
helm repo add stable https://charts.helm.sh/stable
helm repo update

Supportability Metrics

Two major kinds of deployments are provided based on different vSAN versions.

  • From vSAN 7.0, each vSAN host provides vSAN Prometheus endpoints, so the deployed Prometheus instance can easily scrape vSAN Perf data through the endpoints.
  • To support the older vSAN versions pre vSAN 7.0, this repo provides vsan-prometheus-exporter container to translate vSAN Perf data in Prometheus format.

For each deployment, this repo also provides two options either SideCar or Prometheus Operator way. You can follow either option to intergate existing Prometheus monitoring solution.

Some of vSAN Perf metrics are not avaliable for old vSAN versions, which makes the lacking of data in Grafana dashboards.

The following supportability table summarize those difference between vSAN versions

vSAN Version Deployment Solutions Dashboard Graph Unavaliable
vSAN 6.5U3 vSAN Prometheus Exporter vSAN CPU - LLOG, Network, CMMDS; DOM DG, vSAN Layers - DOM CompMgr DG
vSAN 6.7U3 vSAN Prometheus Exporter None
vSAN 7.0 vSAN Prometheus Exporter; Built-In Endpoint None
vSAN 7.0U1 vSAN Prometheus Exporter; Built-In Endpoint None

Deploying

In vSAN 7.0, the vSAN Prometheus endpoint and metrics authorization are introduced in the product, thus the monitoring solutions have difference between vSAN 7.0 and vSAN pre 7.0.

Step 0.1: Please clone this repo and go to the repo folder for all following commands.

git clone https://github.com/vmware/vsan-integration-for-prometheus.git
cd vsan-integration-for-prometheus

Step 0.2: Create Kubernetes namespace and make it default for following kubectl commands

kubectl create namespace vsan-prometheus
kubectl config set-context --current --namespace=vsan-prometheus

vSAN 7.0

Two options are provided for monitoring vSAN 7.0: sidecar container solution and Prometheus Operator. Before applying either option, the metrics authorization token setup and Grafana dashboard configMap should be prepared.

Step 1: vSphere admin requests metrics authorization token for a specific vSAN cluster.

docker run -it vmware/vsan-prometheus-setup:v20210717 --host <vCenter> --username <userName> --cluster <clustername>

This is a sample output from above command. Learn more vsan-prometheus-setup container usage, please go to its readme.

Enter password for host 10.160.29.255 and username [email protected]:
Successfully connected to vCenter!
Successfully generate a new token:
47718bb7-e368-4c01-beed-943cce

Step 2: Create configmap for Grafana dashboards.

kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards

Step 3: It is necessary to label configMaps so that the dashboards are auto imported.

kubectl label configmap grafana-dashboards grafana_dashboard=1

Option 1: Sidercar container

In prometheus-value.yaml, we customize the default Prometheus Helm chart for adding a sideCar container named vsan-prometheus-servicediscovery. This container regularly updates Prometheus server list by querying vSAN service discovery endpoint.

In grafana-value.yaml, we customize the default Grafana Helm chart for reaching the Prometheus server, and adding our vSAN dashboards.

Note:

  • The default Prometheus Helm chart enables metrics for Kubernates cluster and Prometheus, such as kubernetes-apiservers, kubernetes-nodes, kubernetes-pods, etc.
  • The default Grafana Helm chart does not provide any default dashboards. In our customized file, we add basic Grafana charts for Kubernetes cluster metrics.

Step 1: Create secret of bearer token and vCenter for passing them to Prometheus.

kubectl create secret generic bearer-token-secret --from-literal=bearer-token=<token> --from-literal=vcenter=<vCenter>

Note:

  • The bearer token secret is used for passing environment variables for vsan-prometheus-servicediscovery container. More details can be found in this container readme.

Step 2: Install Prometheus:

helm install -f yaml/prometheus-value.yaml prometheus prometheus-community/prometheus

Step 3: Install Grafana:

helm install -f yaml/grafana-value.yaml grafana grafana/grafana

Example commands are provided with a demo video.

kubectl create secret generic bearer-token-secret --from-literal=bearer-token=<token> --from-literal=vcenter=<vCenter>
kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards
kubectl label configmap grafana-dashboards grafana_dashboard=1
helm install -f yaml/prometheus-value.yaml prometheus prometheus-community/prometheus
helm install -f yaml/grafana-value.yaml grafana grafana/grafana

Note: The example run skips vCenter server security verification. To enable vSAN Prometheus security verification, please go to vsan-prometheus-servicediscovery container readme.

Uninstall Helm charts or reinstall the charts

helm delete prometheus grafana

Option 2: Prometheus Operator

In addition to Prometheus operator helm chart, we build an add-on integration for having our own helm chart named vSAN Prometheus operator service discovery The helm chart wraps vsan-prometheus-operator container, which generates serviceMonitor for tracking vSAN Prometheus endpoints. Get more details from this container readme.

Step 1: Install helm chart vSAN Prometheus service discovery:

helm install vsan-operator helm/vsan-prometheus-operator-service-discovery --set vcenter=xxx --set bearerToken=xxx

Step 2: Install helm chart Prometheus Operator:

helm install -f yaml/prometheus-operator-value.yaml vsan-monitor prometheus-community/kube-prometheus-stack

Example commands are provided with a demo video.

kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards
kubectl label configmap grafana-dashboards grafana_dashboard=1
helm install vsan-operator helm/vsan-prometheus-operator-service-discovery --set vcenter=xxx --set bearerToken=xxx
helm install -f yaml/prometheus-operator-value.yaml vsan-monitor prometheus-community/kube-prometheus-stack

Note:

  • The helm chart vsan-prometheus-operator-service-discovery generates auth secret and deploys vsan-prometheus-operator container. And the vsan-prometheus-operator container generates ServiceMonitor object for monitoring vSAN endpoints.
  • Since serviceMonitor monitors related Kubernetes Service and Endpoint, so Prometheus Operator discovers the Endpoint objects and configures Prometheus to monitor them.
  • If security certificate works for IP address in the Endpoint object, please follow helm chart value.yaml to pass certificate file.
  • Please match the secret and serviceMonitor name in prometheus-operator-value.yaml if you change the default values.

Uninstall Helm charts or reinstall the charts

helm delete vsan-operator vsan-monitor

vSAN Pre 7.0

We create our own helm chart for exporting vSAN pre 7.0 Prometheus data. The helm chart deploys vsan-prometheus-exporter container. It creates a Kubernetes service for accessing the container. Get more details from this container readme.

The following steps are to prepare Kubernetes resources for monitoring solution deployment

Step 1: Generate a bearer token since vSAN pre 7.0 does not support it.

export BEARER_TOKEN=$(uuidgen)

Step 2: Create configmap for Grafana dashboards.

kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards

Step 3: It is necessary to label configMaps so that the dashboards are auto imported.

kubectl label configmap grafana-dashboards grafana_dashboard=1

Option 1: Sidercar container

Step 1: Install helm chart vSAN Prometheus exporter:

helm install vsan-exporter helm/vsan-prometheus-exporter --set vcenter=xxx --set username='xxx' --set passwd='xxx' --set bearerToken=$BEARER_TOKEN

Step 2: Install helm chart Prometheus:

helm install -f yaml/prometheus-value.yaml prometheus prometheus-community/prometheus

Step 3 Install helm chart Grafana:

helm install -f yaml/grafana-value.yaml grafana grafana/grafana

Example commands are provided with a demo video.

export BEARER_TOKEN=$(uuidgen)
kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards
kubectl label configmap grafana-dashboards grafana_dashboard=1
helm install vsan-exporter helm/vsan-prometheus-exporter --set vcenter=xxx --set username='xxx' --set passwd='xxx' --set bearerToken=$BEARER_TOKEN
helm install -f yaml/prometheus-value.yaml prometheus prometheus-community/prometheus
helm install -f yaml/grafana-value.yaml grafana grafana/grafana

Uninstall Helm charts or reinstall the charts

helm delete prometheus grafana

Option 2: Prometheus Operator

Step 1: Install helm chart vSAN Prometheus exporter:

helm install vsan-exporter helm/vsan-prometheus-exporter --set vcenter=xxx --set username='xxx' --set passwd='xxx' --set bearerToken=$BEARER_TOKEN

Step 2: Install helm chart Prometheus Operator:

helm install -f yaml/prometheus-operator-value.yaml vsan-monitor prometheus-community/kube-prometheus-stack

Step 3: Install helm chart vSAN operator service discovery:

  • Step 3-1: Get the vcenter service name
export VC_SERVICE=$(kubectl get svc -l "app=vsan-prometheus-exporter" -o jsonpath="{.items[0].metadata.name}:{.items[0].spec.ports[0].port}")
  • Step 3-2: Install helm chart vSAN operator service discovery
helm install vsan-operator helm/vsan-prometheus-operator-service-discovery --set vcenter=$VC_SERVICE  --set bearerToken=$BEARER_TOKEN --set scheme=http

Example commands are provided with a demo video.

export BEARER_TOKEN=$(uuidgen)
kubectl create configmap grafana-dashboards --from-file=grafana-dashboard/dashboards
kubectl label configmap grafana-dashboards grafana_dashboard=1
helm install vsan-exporter helm/vsan-prometheus-exporter --set vcenter=xxx --set username='xxx' --set passwd='xxx' --set bearerToken=$BEARER_TOKEN
helm install -f yaml/prometheus-operator-value.yaml vsan-monitor prometheus-community/kube-prometheus-stack
export VC_SERVICE=$(kubectl get svc -l "app=vsan-prometheus-exporter" -o jsonpath="{.items[0].metadata.name}:{.items[0].spec.ports[0].port}")
helm install vsan-operator helm/vsan-prometheus-operator-service-discovery --set vcenter=$VC_SERVICE --set bearerToken=$BEARER_TOKEN --set scheme=http

Uninstall Helm charts or reinstall the charts

helm delete vsan-exporter vsan-monitor vsan-operator

Accessing

We suggest to install VMware octant, which is an open source tool to understand applications run on a Kubernetes cluster. Using Octant, you can explore Prometheus and Grafana container. By container port forwarding, it gives you local access URLs.

Prometheus & Grafana Url

NodePort for exposing a static port

To get the Prometheus & Grafana access url, you can describe the Kubernetes nodes for having the master node external IP.

kubectl get node -o wide

If the master node external IP is none, you can try your machine IP address, or hostname or using localhost as url.

Option 1: Sidercar container

Grafana

  • Get the Grafana URL to visit by running the command: 
    kubectl port-forward svc/grafana  30000:80
  • URL: http://url:30000/
  • UserName / Password : admin / grafana

Prometheus

  • Get the Prometheus server URL by running these commands in the same shell: 
    kubectl port-forward svc/prometheus-server 30001:80
  • URL: http://url:30001

Option 2: Prometheus Operator

Grafana

  • Instead of using octant, you can apply the command line for port forwarding 
    kubectl port-forward svc/vsan-monitor-grafana 30000:80
  • URL: http://url:30000/
  • UserName / Password : admin / grafana

Prometheus

  • Instead of using octant, you can apply the command line for port forwarding 
    kubectl port-forward svc/vsan-monitor-kube-promethe-prometheus 30001:9090
  • URL: http://url:30001

Grafana dashboard access

Once you login Grafana, you can access the predefined dashboards by clicking the dashboard menu, then choosing Manage choice. Or you append the dashboards after the Grafana URL, e.g. http://localhost:30000/dashboards

Here is a Grafana dashboards list screenshot.

Uninstalling

To remove deployed vSAN Prometheus, you can use the following commands.

Uninstall Helm charts

vSAN 7.0 Uninstall

Sidercar containers

helm delete prometheus grafana

Prometheus Operator

helm delete vsan-operator vsan-monitor

vSAN pre 7.0 Uninstall

Sidercar containers

helm delete prometheus grafana vsan-exporter

Prometheus Operator

helm delete vsan-exporter vsan-monitor vsan-operator

Reset Kubernetes namespace

kubectl delete secret bearer-token-secret 
kubectl delete configmap grafana-dashboards
kubectl config set-context --current --namespace=default
kubectl delete namespace vsan-prometheus

Dashboards

We provide pre-defined Grafana dashboards for monitoring the key metrics of vSAN instead of learning vSAN Prometheus metrics details. The document of Grafana dashboard provides how to organize and generate Grafana dashboards.

Ephemeral deployments for vSAN benchmark

We use Thanos to upload Prometheus snapshots to S3 storage. The benchmark ephemeral document provides the idea, the design and implementation.

Please go to Thanos getting started for having a try!

Development

We provide technical details in development document. You can build your own container images or run testing codes based on your requirements.