Installation on Kubernetes Cluster

Create secrets

Create a secret object for the ATSD collector user. This account will be used by collector instances to transmit metrics securely into the database.

Create a secret.yaml file with the following content:

apiVersion: v1
kind: Secret
metadata:
  name: axibase
type: Opaque
data:
  collector-pass: MTIzNDU2Nzg5MA==

Create the secret using the kubectl create command:

$ kubectl create -f ./secret.yaml
secret "axibase" created

Install ATSD

Create ATSD Deployment

Create a deployment object for ATSD configured using the following atsd-deployment.yaml file as an example.

apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: atsd
  labels:
    app: atsd
    tier: deployment
spec:
  selector:
    matchLabels:
      app: atsd
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        app: atsd
    spec:
      containers:
      - image: axibase/atsd:latest
        name: atsd
        env:
        - name: COLLECTOR_USER_TYPE
          value: api-rw
        - name: COLLECTOR_USER_NAME
          value: collector
        - name: COLLECTOR_USER_PASSWORD
          valueFrom:
            secretKeyRef:
              name: axibase
              key: collector-pass
        ports:
        - containerPort: 8081
          name: tcp
        - containerPort: 8443
          name: https
        resources:
          requests:
            memory: "600Mi"
          limits:
            memory: "1200Mi"

Notes:

atsd deployment is defined in metadata: name field.
The deployment consists of one pod.
The selector field defines how the deployment finds pods to be managed. In this case, we simply select a pod based on the label defined in the Pod template (app: atsd).
The pod template’s specification, or template: spec field, indicates that the Pod should run one container, named atsd, which runs the latest atsd image from Docker Hub.
The deployment opens ports 8081 and 8443.

The template field contains the following instructions for the pod:

Add label app: atsd to the container.
Create one container and name it atsd.
Run a container from the latest axibase/atsd image.
The container should use secrets as environment variables, indicated by valueFrom.secretKeyRef field.
The container has a memory request of 600 MiB and a memory limit of 1200 MiB.
Open ports 8081 and 8443 to accept incoming TCP traffic.

Create the deployment using the kubectl create command:

$ kubectl create -f ./atsd-deployment.yaml
deployment "atsd" created

Create ATSD Service

Create a service object for ATSD using atsd-service.yaml file with the following content:

apiVersion: v1
kind: Service
metadata:
  name: atsd
  labels:
    app: atsd
    tier: service
spec:
  ports:
    - protocol: TCP
      port: 8081
      nodePort: 30081
      name: tcp
    - protocol: TCP
      port: 8443
      nodePort: 30443
      name: https
  selector:
    app: atsd
  type: NodePort

This specification will create a new Service object named atsd which opens TCP ports on any Pod with the app=atsd label.

The service type is specified as NodePort, whereas each node will redirect traffic to the atsd service on the same port. The Service automatically creates environment variables, which are supported by Collector:

ATSD_SERVICE_HOST (variable pattern '{service_name}_SERVICE_HOST')
ATSD_SERVICE_PORT_HTTPS (variable pattern '{service_name}_SERVICE_PORT_{port_name}'), specified by the spec.ports field:
```
spec:
  ports:
    - protocol: TCP
      port: 8443
      name: https
```
ATSD_SERVICE_PORT_TCP (variable pattern '{service_name}_SERVICE_PORT_{port_name}'), specified by the spec.ports field:
```
spec:
  ports:
    - protocol: TCP
      port: 8081
      name: tcp
```

Create the service using the kubectl create command:

$ kubectl create -f ./atsd-service.yaml
service "atsd" created

Note When using type NodePort, the firewall is not opened by default. You need to add firewall rules to allow access for ports, indicated in the nodePort: field.

Install Collector

Create Collector Deployment

Create a deployment object for Axibase Collector using collector-deployement.yaml file:

apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: axibase-collector
  labels:
    app: axibase-collector
    tier: deployment
spec:
  replicas: 3
  selector:
    matchLabels:
      app: axibase-collector
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        app: axibase-collector
    spec:
      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
          - labelSelector:
             matchExpressions:
             - key: app
               operator: In
               values:
               - axibase-collector
            topologyKey: kubernetes.io/hostname
      containers:
      - image: axibase/collector:latest
        name: axibase-collector
        env:
        - name: COLLECTOR_USER_NAME
          value: collector
        - name: COLLECTOR_USER_PASSWORD
          valueFrom:
            secretKeyRef:
              name: axibase
              key: collector-pass
        - name: JAVA_OPTS
          value: "-Xmx512m"
        args: ["-job-enable=docker-socket"]
        ports:
        - containerPort: 9443
          hostPort: 9443
          name: https
        volumeMounts:
        - name: docker-socket
          mountPath: /var/run/docker.sock
        resources:
          requests:
            memory: "200Mi"
          limits:
            memory: "600Mi"
      volumes:
      - name: docker-socket
        hostPath:
          path: /var/run/docker.sock

Notes:

axibase-collector deployment is defined in metadata: name field.
The Deployment creates three replicated Pods, indicated by the replicas field. If you want to install the Collector on each node, the replicas value must be equal node count.
The selector field defines how the deployment finds pods to be managed. In this case, we select a pod based on the label defined in the Pod template (app: axibase-collector).
The pod template’s specification, or template: spec field, indicates that the Pod should run one container, named axibase-collector, which runs the latest axibase-collector image from Docker Hub.
The deployment opens port 9443.

The template field contains the following instructions:

Add label app: axibase-collector to each pod.
Pods use affinity. We have podAntiAffinity that will restrict to have two same pods (label app:axibase-collector) in one node.
Create one container and name it axibase-collector.
Create one volume and name it docker-socket.
Run the axibase/collector image at latest version.
The container uses secrets as environment variables, indicated by valueFrom.secretKeyRef field.
The container has a memory request of 200 MiB and a memory limit of 600 MiB.
The container uses the docker-socket volume, indicated by the volumeMounts field.
Open ports 9443 so that the container can send and accept traffic. The port is also opened on each Node, specified by the hostPort: 9443 field.

Create the deployment using the kubectl create command:

$ kubectl create -f ./collector-deployment.yaml
deployment "axibase-collector" created

Verify Installation

View deployments using kubectl get:

$ kubectl get deploy -o wide
NAME                DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE       CONTAINERS          IMAGES                     SELECTOR
atsd                1         1         1            1           2d        atsd                axibase/atsd:latest        app=atsd
axibase-collector   3         3         3            3           2d        axibase-collector   axibase/collector:latest   app=axibase-collector

View pods using kubectl get:

$ kubectl get pod -o wide
NAME                                 READY     STATUS    RESTARTS   AGE       IP          NODE
atsd-1742852640-vvzlk                1/1       Running   0          2d        10.8.0.14   gke-cluster-1-default-pool-fcd89b74-bstn
axibase-collector-2747957227-03dj2   1/1       Running   0          2d        10.8.2.8    gke-cluster-1-default-pool-fcd89b74-96gq
axibase-collector-2747957227-5hr0g   1/1       Running   0          2d        10.8.0.15   gke-cluster-1-default-pool-fcd89b74-bstn
axibase-collector-2747957227-whd99   1/1       Running   0          2d        10.8.1.8    gke-cluster-1-default-pool-fcd89b74-ln58

View services using kubectl get:

$ kubectl get service -o wide
NAME         TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                         AGE       SELECTOR
atsd         NodePort    10.11.241.144   <none>        8081:30081/TCP,8443:30443/TCP   3d        app=atsd
kubernetes   ClusterIP   10.11.240.1     <none>        443/TCP                         3d        <none>

# Installation on Kubernetes Cluster

# Create secrets

# Install ATSD

# Create ATSD Deployment

# Create ATSD Service

# Install Collector

# Create Collector Deployment

# Verify Installation