Skip to content
This repository has been archived by the owner on Jan 12, 2021. It is now read-only.

Latest commit

 

History

History
243 lines (186 loc) · 12.2 KB

README.md

File metadata and controls

243 lines (186 loc) · 12.2 KB
Raw

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.

Prerequisites

Deployment using ClusterManagerTemplate.yaml

  1. Create a yaml deployment file using ClusterManagerTemplate.yaml. For example, you can use a bash script to set the required environment variables and generate a new file called ClusterManager.yaml.

    #!/bin/bash

# Set environment variables 
export FEDERATION_HOST=fedhost
export FEDERATION_CONTEXT=akube
export FEDERATION_NAMESPACE=federation-system
export IMAGE_REPOSITORY=someregistry.io/somewhere
export IMAGE_VERSION=tagversion
export DOMAIN=something.net
export KOPS_STATE_STORE=s3://state-store.something.net
export AWS_ACCESS_KEY_ID=awsaccesskeyid
export AWS_SECRET_ACCESS_KEY=awssecretaccesskey
export OKE_BEARER_TOKEN=werckerclustersbearertoken
export OKE_AUTH_GROUP=werckerclustersauthgroup
export OKE_CLOUD_AUTH_ID=werckerclusterscloudauthid
 
# Convert API access keys to base64 for Kubernetes secret storage.
export AWS_ACCESS_KEY_ID_BASE64=$(echo -n "$AWS_ACCESS_KEY_ID" | base64)
export AWS_SECRET_ACCESS_KEY_BASE64=$(echo -n "$AWS_SECRET_ACCESS_KEY" | base64)
export OKE_BEARER_TOKEN_BASE64=$(echo -n "$OKE_BEARER_TOKEN" | base64)
export OKE_AUTH_GROUP_BASE64=$(echo -n "$OKE_AUTH_GROUP" | base64)
export OKE_CLOUD_AUTH_ID_BASE64=$(echo -n "$OKE_CLOUD_AUTH_ID" | base64)
 
# Generate a new ClusterManager.yaml by using ClusterManagerTemplate.yaml and environment variables.
eval "cat <<EOF
$(<ClusterManagerTemplate.yaml)
EOF" > ClusterManager.yaml

  2. Use kubectl to deploy the generated ClusterManager.yaml file.

    kubectl --context $FEDERATION_HOST create -f ClusterManager.yaml

  3. Verify if Cluster Manager is installed.

    kubectl --context $FEDERATION_HOST get pods --all-namespaces | grep cluster-manager

  4. (Optional) Uninstall Cluster Manager.

    kubectl --context $FEDERATION_HOST delete -f ClusterManager.yaml

Deployment using Helm

  1. Install tiller if not installed on the Federation host.

    helm init

  2. Execute helm install with the following values set. Note that the following command makes a reference to helm chart in the project's deploy directory (./helm/cluster-manager). Therefore this command should either run from the deploy directory or may refer to the helm chart in the deploy directory.

    FEDERATION_NAMESPACE=federation-system
FEDERATION_HOST=fedhost
FEDERATION_CONTEXT=akube
helm install ./helm/cluster-manager \
    --name cluster-manager \
    --set awsAccessKeyId="AWSACCESSKEYID" \
    --set awsSecretAccessKey="AWSSECRETACCESSKEY" \
    --set okeBearerToken="WerckerClustersBearerToken" \
    --set okeAuthGroup="WerckerClustersAuthGroup" \
    --set okeCloudAuthId="WerckerClustersCloudAuthId" \
    --set federationEndpoint="https://$FEDERATION_CONTEXT-apiserver" \
    --set federationContext="$FEDERATION_CONTEXT" \
    --set federationNamespace="$FEDERATION_NAMESPACE" \
    --set domain="something.fed.net" \
    --set image.repository="someregistry.io/somewhere" \
    --set image.tag="v1" \
    --set okeApiHost="api.cluster.us-ashburn-1.oracledx.com" \
    --set statestore="s3://clusters-state" \
    --kube-context "$FEDERATION_HOST"

    Where,

    • FEDERATION_NAMESPACE - Namespace where federation was installed via kubefed init
    • FEDERATION_HOST - Federation host name
    • FEDERATION_CONTEXT - Federation context name
    • awsAccessKeyId - AWS access key ID
    • awsSecretAccessKey - AWS secret access ID
    • okeBearerToken- Wercker Clusters bearer token
    • okeAuthGroup - Wercker Clusters auth group
    • okeCloudAuthId - Wercker Clusters cloud auth ID
    • federationEndpoint - Federation API server endpoint
    • federationContext - Federation name
    • federationNamespace - Similar to FEDERATION_NAMESPACE. Defaults to federation-system if not set.
    • domain - AWS domain name
    • image.repository - Repository where Cluster Manager is stored
    • image.tag - Cluster Manager image tag/version no.
    • okeApiHost - Wercker Clusters API host endpoint
    • statestore - AWS cluster S3 state store

  3. Verify if Cluster Manager is installed.

    helm --kube-context $FEDERATION_HOST list cluster-manager

  4. (Optional) Uninstall Cluster Manager.

    helm delete --kube-context $FEDERATION_HOST --purge cluster-manager

Examples on how to use Cluster Manager

Provisioning a Wercker Cluster

  1. Deploy a Wercker Cluster to the Federation, use ClusterOke.yaml from examples.

    kubectl --context $FEDERATION_CONTEXT create -f ClusterOke.yaml

    This command deploys the cluster in an offline state. To customize the Wercker cluster, change the parameters in cluster-manager.n6s.io/cluster.config annotation of cluster. Before deploying, you can modify ClusterOke.yaml or use kubectl annotate command. The supported parameters are

    • K8Version - Kubernetes Version. Valid values are 1.7.4, 1.7.9, or 1.8.0. Default value is 1.7.4.
    • nodeZones - Available Domains. Valid values are AD-1, AD-2, AD-3. This field can be set to any combination of the set of values.
    • shape - Valid values are VM.Standard1.1, VM.Standard1.2, VM.Standard1.4, VM.Standard1.8, VM.Standard1.16, VM.DenseIO1.4, VM.DenseIO1.8, VM.DenseIO1.16, BM.Standard1.36, BM.HighIO1.36 and BM.DenseIO1.36.
    • workersPerAD - No. of worker nodes per availability domain.
    • compartment - Wercker Cluster Compartment ID where worker nodes will be instantiated.

  2. Initiate provisioning. Note: Do not perform these steps if you are using Navarkos. Navarkos determines on which cluster to perform the operation depending on demand and supply.

    • Update annotation n6s.io/cluster.lifecycle.state with value pending-provision.

    • At the end of the provisioning, the value of n6s.io/cluster.lifecycle.state will be changed either to ready if successful or failed-up if failed.

      kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-oke n6s.io/cluster.lifecycle.state=pending-provision --overwrite

Provisioning an AWS Cluster

  1. Deploy an AWS cluster to the Federation using KOPS, use ClusterAwsEast2.yaml from examples.

    kubectl --context $FEDERATION_CONTEXT create -f ClusterAwsEast2.yaml

    This command deploys the cluster in an offline state. To customize the AWS cluster, change the parameters in cluster-manager.n6s.io/cluster.config. For valid values that are going to be used on some of the fields in cluster-manager.n6s.io/cluster.config which is used to customize the cluster, please refer to the AWS Documentation:

    • region - Region where the cluster will be instantiated.
    • masterZones - List of zones where master nodes will be instantiated.
    • nodeZones - List of zones where worker nodes will be instantiated.
    • masterSize - Master node(s) instance size
    • nodeSize - Worker node(s) instance size.
    • numberOfMasters - No. of master nodes to instantiate.
    • numberOfNodes - No. of worker nodes to instantiate.
    • sshkey - SSH public access key to the cluster nodes.

  2. Initiate provisioning. Note: Do not perform this step if you are using Navarkos. Navarkos determines on which cluster to perform the operation depending on demand and supply.

    • Update annotation n6s.io/cluster.lifecycle.state with value pending-provision.
    • At the end of the provisioning, the value of n6s.io/cluster.lifecycle.state will be changed either to ready if successful or failed-up if failed.
    kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 n6s.io/cluster.lifecycle.state=pending-provision --overwrite

Scaling up an already provisioned cluster

Note: Do not perform these steps if you are using Navarkos. Navarkos determines on which cluster to perform the operation depending on demand and supply.

  • If there is more demand, you can scale up an already provisioned cluster to support more load. Update the annotation n6s.io/cluster.lifecycle.state with value pending-up.
  • You can configure the scale up size using the annotation cluster-manager.n6s.io/cluster.scale-up-size, otherwise it uses the default value 1.
  • At the end of the provisioning, the value of n6s.io/cluster.lifecycle.state will be changed either to ready if successful or failed-up if failed.

Here is an example of scaling up a previously provisioned akube-us-east-2 AWS cluster by 5

kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 cluster-manager.n6s.io/cluster.scale-up-size=5 --overwrite && \
kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 n6s.io/cluster.lifecycle.state=pending-up --overwrite

Scaling down an already provisioned cluster

Note: Do not perform these steps if you are using Navarkos. Navarkos determines on which cluster to perform the operation depending on demand and supply.

  • If the demand is low, then you can scale down an already provisioned cluster. Update the annotation n6s.io/cluster.lifecycle.state with value pending-down.
  • You can configure scale down size using the annotation cluster-manager.n6s.io/cluster.scale-down-size, otherwise it uses the default value 1.
  • At the end of the provisioning, the value of n6s.io/cluster.lifecycle.state will be changed either to ready if successful or failed-up if failed.

Here is an example of scaling down a previously provisioned akube-us-east-2 AWS cluster by 5

kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 cluster-manager.n6s.io/cluster.scale-down-size=5 --overwrite && \
kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 n6s.io/cluster.lifecycle.state=pending-down --overwrite

Shutting down a provisioned cluster

Note: Do not perform these steps if you are using Navarkos. Navarkos determines on which cluster to perform the operation depending on the demand and supply.

  • You can shut down a provisioned cluster when it is not in use. Update annotation n6s.io/cluster.lifecycle.state with value pending-shutdown.
  • At the end of the provisioning, the value of n6s.io/cluster.lifecycle.state will be changed to offline.

Here is an example of shutting down a previously provisioned akube-us-east-2 AWS cluster

kubectl --context=$FEDERATION_CONTEXT annotate cluster akube-us-east-2 n6s.io/cluster.lifecycle.state=pending-shutdown --overwrite

Checking cluster status

  • To check the status of a specific cluster, execute the command and note the value of n6s.io/cluster.lifecycle.state annotation.

    kubectl --context=$FEDERATION_CONTEXT get cluster <Cluster Name> -o yaml

  • To check the status of all the clusters, execute the command and note the value of n6s.io/cluster.lifecycle.state annotation for each of the clusters listed .

    kubectl --context=$FEDERATION_CONTEXT get clusters -o yaml