Skip to content

Latest commit

 

History

History
78 lines (53 loc) · 2.45 KB

k8s-deploy.md

File metadata and controls

78 lines (53 loc) · 2.45 KB
Raw

Deploying in Kubernetes with Oxia cluster Helm chart

Deploying the Oxia cluster

To deploy the Oxia cluster with Helm:

$ kubectl create namespace oxia

$ git clone https://github.com/streamnative/oxia.git

$ cd oxia

$ helm upgrade --install oxia \
  --namespace oxia \
  --set image.repository=streamnative/oxia \
  --set image.tag=main \
  --set image.pullPolicy=IfNotPresent \
  deploy/charts/oxia-cluster

Monitoring Oxia

Oxia support monitoring through exposing a ServiceMonitor profile. If you have already a Prometheus deployment int your Kubernetes cluster, you won't need any extra steps. It would directly start collecting monitoring data from Oxia clusters.

The Helm Chart for the Oxia cluster uses a monitoringEnabled flag to decide whether to install the service monitor. If you don't have Prometheus installed and don't want to install it, you can set monitoringEnabled: false to skip this part.

Grafana's dashboards are available at deploy/dashboards. These can just be imported in your existing Grafana instance.

Deploying Prometheus Stack

If you don't have already a Prometheus deployment, you can easily add it with Helm.

Prepare:

$ kubectl create namespace monitoring

$ helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
$ helm repo update

Deploying the stock Prometheus & Grafana

This installs the stock Prometheus & Grafana, without the Oxia dashboards, which you will be able to manually import later.

$ helm install monitoring prometheus-community/kube-prometheus-stack \
          --namespace monitoring \
          --set prometheus.prometheusSpec.serviceMonitorSelectorNilUsesHelmValues=false

Deploying Prometheus & Grafana with Oxia dashboards

To deploy the stack with Oxia's Grafana dashboards pre-configured:

Note: The unpack method has to be used as the grafana chart utilises Helm's .Files.Get to load custom dashboards which cannot reach files outside the chart directory.

$ helm pull prometheus-community/kube-prometheus-stack
$ tar -xf kube-prometheus-stack-*.tgz
$ cp deploy/dashboards/*.json kube-prometheus-stack/charts/grafana/dashboards

$ helm upgrade --install monitoring kube-prometheus-stack \
  --namespace monitoring \
  --values deploy/dashboards/values-kube-prometheus-stack.yaml

$ rm -rf kube-prometheus-stack*

Note: The default login credentials for grafana are admin/prom-operator.