Deploy Aerospike Graph Service with Kubernetes

This page describes how to run Aerospike Graph Service (AGS) using Kubernetes and connect to an Aerospike Database cluster.

Aerospike Database requirements

• A running Aerospike Database deployment, version 7.0 or later.
• If access control is enabled on the Aerospike Database, AGS must have sys-admin and read-write privileges as a database user.
• Verify the namespace AGS uses on your Aerospike Database server has the configuration option default-ttl set to 0. See the TTL configuration instructions for setting this option after the server has started.
• If TLS is required between AGS and the Aerospike Database, enable it and provide the TLS name and port via Helm parameters. See TLS between AGS and Aerospike DB for configuration details.

Deploy AGS on Kubernetes

Select a deployment platform from the following tabs.

Local (minikube)

Deploying locally requires minikube, kubectl, and Helm.

Deploy AGS on minikube

Tip

Skip to step 3 if you already have Aerospike Database deployed in minikube.

1. Start a Kubernetes cluster locally using minikube:

   minikube start

2. Deploy Aerospike Database on minikube.

   See Install the Aerospike Kubernetes Operator to install AKO using Helm, then go to Deploy Aerospike with Helm to deploy Aerospike on the cluster.

   Note

   Persistent storage is not required for local testing, but any data stored in Aerospike will be temporary and erased after restarts.

3. Verify you are pointing to the right Kubernetes context:

   kubectl config get-contexts

   The current context should point to your minikube cluster.

Amazon EKS

Deploying on EKS requires eksctl, kubectl, and Helm.

Deploy AGS on EKS

1. Provision an EKS cluster.

   Create an EKS cluster in the same region and availability zone as your Aerospike Database. Replace the placeholder values with your AWS configuration:

   ags-eks-cluster.yaml

   apiVersion: eksctl.io/v1alpha5
   kind: ClusterConfig
   metadata:
     name: ags-eks-cluster
     region: YOUR_REGION
   vpc:
     id: YOUR_VPC_ID
     subnets:
       public:
         YOUR_AVAILABILITY_ZONE_A:
           id: YOUR_SUBNET_ID_A
         YOUR_AVAILABILITY_ZONE_B:
           id: YOUR_SUBNET_ID_B
         YOUR_AVAILABILITY_ZONE_C:
           id: YOUR_SUBNET_ID_C
   iam:
     withOIDC: true
   managedNodeGroups:
     - name: ng-ags-test
       instanceType: m5d.2xlarge
       desiredCapacity: 2
       minSize: 1
       maxSize: 3

   Run via eksctl:

   eksctl create cluster -f ags-eks-cluster.yaml

2. Verify you are pointing to the right Kubernetes context:

   kubectl config get-contexts

   The current context should point to your EKS cluster.

3. Verify the EKS cluster is ready:

   kubectl get nodes

   You should see your EKS worker nodes in Ready state.

Google GKE

Deploying on GKE requires gcloud CLI, kubectl, and Helm.

Deploy AGS on GKE

1. Provision a GKE cluster.

   Create a GKE cluster in the same region and availability zone as your Aerospike Database:

   gcloud container clusters create aerospike-graph1 \
     --zone us-central1-a \
     --num-nodes 3 \
     --machine-type e2-standard-4 \
     --enable-ip-alias \
     --no-enable-autoupgrade \
     --no-enable-autorepair

2. Verify you are pointing to the right Kubernetes context:

   kubectl config get-contexts

   The current context should point to your GKE cluster.

3. Verify the GKE cluster is ready:

   kubectl get nodes

   You should see your GKE worker nodes in Ready state:

   NAME                                              STATUS   ROLES    AGE    VERSION
   gke-aerospike-graph1-default-pool-3da299ff-7qc1   Ready    <none>   116s   v1.32.4-gke.1415000
   gke-aerospike-graph1-default-pool-3da299ff-vp25   Ready    <none>   116s   v1.32.4-gke.1415000
   gke-aerospike-graph1-default-pool-3da299ff-x9t7   Ready    <none>   116s   v1.32.4-gke.1415000

Install AGS with Helm

After setting up your Kubernetes cluster, install AGS using Helm.

1. Add the Aerospike Helm repository:

   helm repo add aerospike https://artifact.aerospike.io/helm

2. Update the Helm repository:

   helm repo update

3. Deploy AGS using Helm.

   Replace the placeholder values with your Aerospike Database connection details:

   • AEROSPIKE_HOST: Aerospike Database IP address or hostname
   • AEROSPIKE_PORT: Aerospike Database port (default 3000)
   • AEROSPIKE_NAMESPACE: Namespace for graph data (for example, test)
   • AEROSPIKE_USERNAME: (optional) Aerospike Database username if access control is enabled
   • AEROSPIKE_PASSWORD: (optional) Aerospike Database password if access control is enabled
   minikube users

   If you deployed Aerospike Database locally on minikube, get the IP address and port by running kubectl get svc -n aerospike.

   helm install ags aerospike/aerospike-graph \
     --set 'env[0].name=aerospike.client.host' \
     --set 'env[0].value=AEROSPIKE_HOST:AEROSPIKE_PORT' \
     --set 'env[1].name=aerospike.client.namespace' \
     --set 'env[1].value=AEROSPIKE_NAMESPACE' \
     --set 'env[2].name=aerospike.client.user' \
     --set 'env[2].value=AEROSPIKE_USERNAME' \
     --set 'env[3].name=aerospike.client.password' \
     --set 'env[3].value=AEROSPIKE_PASSWORD'

   For TLS configuration, see TLS between AGS and Aerospike DB.

Check AGS pod status

After deploying AGS, verify the pod is running and check its logs.

1. Check the status of the Aerospike Graph pod:

   kubectl get pods

   The status should be Running:

   NAME                                         READY   STATUS    RESTARTS   AGE
   ags-aerospike-graph-589757dd9c-d794m         1/1     Running   0          23m

2. Check the logs to verify that AGS initialized successfully.

   Replace AEROSPIKE_GRAPH_POD_NAME with the pod name from the previous step:

   kubectl logs AEROSPIKE_GRAPH_POD_NAME

3. Get the external IP for connecting to AGS.

   The Aerospike Graph Helm chart deploys a Kubernetes Load Balancer Service by default. Its external IP will be used to connect to AGS in the next section. Run kubectl get svc to check its external IP:

   kubectl get svc

   External IP

   • minikube: Run minikube tunnel in a separate terminal to assign an External IP for the LoadBalancer.
   • Amazon EKS: The External IP is a DNS hostname, for example aadecc03d1c5c42319551ccc4dfe540b-501644508.eu-central-1.elb.amazonaws.com
   • Google GKE: The External IP is a standard IP address, for example 34.61.38.211

Verify the deployment

Gremlin Console is a command-line interface for executing Gremlin traversals against your running graph service.

1. Download and extract the latest Gremlin Console from the Apache TinkerPop downloads:

   curl -O https://dlcdn.apache.org/tinkerpop/3.8.0/apache-tinkerpop-gremlin-console-3.8.0-bin.zip && \
   unzip apache-tinkerpop-gremlin-console-3.8.0-bin.zip && \
   rm apache-tinkerpop-gremlin-console-3.8.0-bin.zip

2. Start the console from the extracted directory:

   apache-tinkerpop-gremlin-console-3.8.0/bin/gremlin.sh

3. Connect to the Aerospike Graph Service instance. Replace EXTERNAL_IP with the external IP from the previous section:

   gremlin> g = traversal().withRemote(DriverRemoteConnection.using("EXTERNAL_IP", 8182, "g"));

   Example response

   ==>graphtraversalsource[emptygraph[empty], standard]

4. Once you are connected via Gremlin Console and it shows gremlin>, you can create and read a record to make sure that everything works properly:

   Run the following command to create a vertex that represents a person:

   gremlin> g.addV('person').property('name', 'Alice').property('age', 29)

   Example response

   ==>v[-1]

   List all vertices and their properties and see the newly created vertex:

   gremlin> g.V().elementMap()

   Example response

   ==>[id:-1,label:person,name:Alice,age:29]