Quickstart: Typesense Operator Installation

Requirements

  • Typesense Operator v2.0 requires Kubernetes 1.20 or later.
  • Works on self-managed and cloud managed clusters

Installation Steps

For an alternative installation method, refer to the TKO Helm Charts.

Set Up Kubernetes Cluster

  • Use minikube or any other tool to create your Kubernetes cluster.

Deploy Custom Resource Definition

  • Navigate to the deployment directory
cd deploy
  • Deploy the Custom Resource Definition
kubectl create -f crd.yaml -n <namespace>

Deploy RBAC (Role-Based Access Control)

  • Deploy RBAC configuration:
kubectl create -f rbac.yaml -n <namespace>

Create secret for apikey used by Typesense

  • Make sure to deploy the secret in the same namespace as the operator:
  • NOTE: do not change key apikey as it is used by operator to identify the value.
kubectl create secret generic -n <namespace> typesense-apikey --from-literal=apikey=MWYyZDFlMmU2N2Rm

Deploy the Latest Operator

  • Deploy the latest Typesense Operator:
kubectl create -f operator.yaml -n <namespace>

Verify Operator Deployment

  • Check if the operator pods are up and running:
kubectl get pods -n <namespace>

Configure Operator

  • Once the operator is running, configure it by referring to the configuration section below.

Operator Configuration

  • Sample configuration can be found in operator-config.yaml
  • All the configuration must be specified under spec or config
  • Supported configurations under spec
    • replicas : Number of replicas of typesense.
      • default : 3
        NOTE: When the number of replicas are increased the operator automatically handles the peering connection between the replicas.
    • namespace: Namespace to be used for deployment.
      • default : typesense
    • image : Typesense production ready docker image. List can be found here.
      • default : latest tag will be pulled
    • resources : Resources for cpu and memory.
      • default
        resources: requests: memory: 100Mi cpu: "64m" limits: memory: 256Mi cpu: "512m"
    • nodeSelector : Node to which pod has to be scheduled. if not specified, it picks up the node with available resources
    • storageClass : Provides a way to administor the storage. Create a storage class with the provider of your choice and add it to the operator config.
      Options available are name and size
      • name : Name of the storageClass that the operator should consider for volume mount.
      • size : Size of the volume to be allocated to each typesense replica.
        NOTE: Supports all k8s Storageclass.
    • startupProbe : Protect slow starting containers with startup probes
      Options available are failureThreshold and periodSeconds
      NOTE: k8s startupProbe.
    • livenessProbe : Kubernetes provides liveness probes to detect and remedy such situations.
      Options available are failureThreshold and periodSeconds
      NOTE: k8s livenessProbe.
  • Supported configurations under config
    • secret : Typesense authentication is done using a password which is stored as secret. Provide the name of the secret here.
      • default : No default value
    • env : Support for supplying environment variables to Typesense. Refer here.
      NOTE: Specifying these env will either add or override existing variables.
      • default : No default value

  • Deploying the configuration

    • Create the operator kubectl create -f operator-config.yaml
    • To apply any changes made to the config kubectl apply -f operator-config.yaml Monitor Cluster Creation

  • Monitor the creation of the Typesense cluster by checking pod statuses:

kubectl get pods -n <namespace>

Operator metrics

  • Operator exposes the following Prometheus metrics to periodically poll this endpoint and export the data into Prometheus
Metrics Description
tok_resource_create_success_total Total number of successful resource creation events by the operator
tok_resource_update_success_total Total number of updates to the resource by the operator
tok_resource_delete_success_total Total number of deletion events by the operator
tok_version_info Idicates the operator version that is currently deployed

Upgrade

  • Before performing this upgrade process, it is recommended to
    • Test it out in a dev environment
    • Backup the Typesense data
  • The process of updating TKO is simple
    1. Watch for changes in the CRD file and apply those changes
    2. Watch for changes in the RBAC file and apply those changes
    3. Apply the new operator.yaml file
    4. Once the new operator pod is up and running, re-apply the configuration of the your operator so that the changes (if any) are reflected.
      kubectl apply -f <your operator config file yaml>

Cleanup

  • When you want to fully remove the cluster operator and associated definitions, you can run
    kubectl delete -f operator-config.yaml
    kubectl delete -f deploy/crd.yaml
    kubectl delete -f deploy/operator.yaml
    kubectl delete -f deploy/rbac.yaml