Skip to main content

Cluster Management

This section describes the process of deploying a Kubernetes cluster via Astro platform on your AWS or GCP account.

Introduction​

There are two types of Kubernetes clusters that you can deploy on your AWS or GCP account:

  • Self-Hosted K8s Clusters: These are Vanilla Kubernetes clusters that Astro platform will manage on your behalf.
  • Managed K8s Clusters: These are Kubernetes clusters that are deployed and managed by Astro platform but uses the cloud provider's managed service for the control plane.

More specifically, you can choose between the following cloud providers:

  • AWS: Deploy on your own VPC or use the VPC created by Astro on your account.
  • GCP: Deploy on your own VPC or use the VPC created by Astro on your account.

Self-Hosted Clusters​

This is vanilla kubernetes cluster that astro platform will manage on your behalf.

AWS​

This section describes the process of deploying a Kubernetes cluster on your AWS account.

API Specification​
Cluster API Specification
clusterName: aws-k8s-dev # name of the cluster
provider: aws # provider of the cluster
notification: # optional, notification will be sent to the slack channel
  type: slack
  slack: 
    webhookURL: https://hooks.slack.com/services/XXXXX # your slack webhook url
region: us-west-2 # region of the cluster
provisioner:
  type: selfHosted # type of the provisioner
  selfHosted:
    accountId: <your-aws-account-id> # your aws account id
    # networkId:  vpc-018f2dec8bb5ddfdd # bring your own VPC
    bucketName: <bucket-name> # bucket name to store the cluster configuration
    credentials: # credentials to access the cluster
      type: static # type of the credentials
      data: 
        AWS_ACCESS_KEY_ID: <your-aws-access-key-id> # your aws access key id
        AWS_SECRET_ACCESS_KEY: <your-aws-secret-access-key> # your aws secret access key
clusterSpec:
  dataPlane: # specification of the data plane
    nodeGroups: # specification of the node groups
    - name: test-mg # name of the node group
      minNode: 3 # minimum number of nodes, 3 is good default given we use three availability zones
      maxNode: 6 # maximum number of nodes
      machineTypes: # machine types of the nodes
      - t3.large # machine type of the nodes
      labels: # labels of the nodes
        foo: "bar"
  controlPlane: # specification of the control plane
    nodeGroup:
      name: control-plane
      machineTypes:
      - t3.medium
Prerequisites

Make sure to follow the prerequisites before you start.

For more information about the cluster specification, see

Operations​

GCP​

This section describes the process of deploying a Kubernetes cluster on your GCP account.

API Specification​
Cluster API Specification
clusterName: foo-cluster # name of the cluster
provider: gcp # provider of the cluster
notification: # optional, notification will be sent to the slack channel
  type: slack # type of the notification
  slack: 
    webhookURL: https://hooks.slack.com/services/xxxx # your slack webhook url
region: us-west2 # region of the cluster
provisioner: 
  type: selfHosted # type of the provisioner
  selfHosted:
    # this is the gcp project id
    accountId: xxxx # your gcp project id
    # networkId:  vpc-018f2dec8bb5ddfdd
    bucketName: <bucket-name> # bucket name to store the cluster configuration
    credentials: # credentials to access the cluster
      type: static # type of the credentials
      data:
        GOOGLE_APPLICATION_CREDENTIALS: <path-to-your-gcp-credentials> # path to your gcp credentials
clusterSpec:
  dataPlane: # specification of the data plane
    nodeGroups: # specification of the node groups
    - name: test-mg # name of the node group
      minNode: 3 # minimum number of nodes, 3 is good default given we use three availability zones
      maxNode: 6 # maximum number of nodes
      instanceType: spot # ondemand or spot
      machineTypes:
      - e2-medium # machine type of the nodes
      labels: # labels of the nodes
        hello: "test"
  controlPlane: # specification of the control plane
    nodeGroup: # specification of the node group
      name: control-plane # name of the node group
      machineTypes: # machine type of the nodes
      - e2-medium # machine type of the nodes\

For more information about the cluster specification, see

Operations​

Managed Clusters​

This is the version of Kubernetes cluster that is deployed and managed by Astro platform but uses the cloud provider's managed service for the control plane.

EKS​

This section describes the process of deploying a Kubernetes cluster on your AWS account.

API Specification​
Cluster API Specification
clusterName: <cluster-name> # name of the cluster
provider: aws # provider of the cluster
notification: # optional, notification will be sent to the slack channel
  type: slack # type of the notification
  slack: 
    webhookURL: https://hooks.slack.com/services/xxxx # your slack webhook url
region: us-west-2 # region of the cluster
provisioner:
  type: eks # type of the provisioner
  eks:
    accountId: <your-aws-account-id> # your aws account id
    credentials: # credentials to access the cluster
      type: dynamic # type of the credentials
#approvalWorkflow: true
clusterSpec: # specification of the cluster
  dataPlane: # specification of the data plane
    nodeGroups: # specification of the node groups
    - name: test-node-group
      minNode: 3 # minimum number of nodes, 3 is good default given we use three availability zones
      maxNode: 6 # maximum number of nodes
      machineTypes: # machine types of the nodes
      - t3.medium
      labels: # labels of the nodes
        hello: test
Prerequisites

Make sure to follow the prerequisites before you start.

Please refer to the following documents for more information about the cluster specification:

Operations​

Cluster Operations​

These are the operations that you can perform on the cluster.

🚀 Cluster Operations​

These are the operations that you can perform on the cluster.

📈 Apply Cluster Configuration​

To apply the updated cluster configuration, execute the following command (admin role required):

astroctl clusters apply -f <cluster-config.yaml>

🗑️ Delete Cluster​

To delete a cluster, execute the following command (admin role required):

astroctl clusters delete <cluster-name>

🛠️ Get Cluster Details​

To get the details of a cluster, execute the following command (no admin required):

astroctl clusters get <cluster-name>

🔑 Manage Kubeconfig​

To generate the kubeconfig file for a cluster, execute the following command (no admin required):

astroctl clusters generate-kubeconfig <cluster-name>

To set the context for a cluster, execute the following command (no admin required):

astroctl clusters set-context <cluster-name>

📊 Check Cluster Status​

This command will show the status of the cluster.

astroctl clusters get <cluster-name>

Note The status subcommand is not available yet!. If your cluster is stuck in INPROGRESS, Please contact Astro platform team for help.

🔄 Upgrade Cluster​

TBD