First steps with Constellation
The following steps guide you through the process of creating a cluster and deploying a sample app. This example assumes that you have successfully installed and set up Constellation.
Create a cluster
-
Create the configuration file for your selected cloud provider.
- Azure
- GCP
constellation config generate azure
constellation config generate gcp
This creates the file
constellation-conf.yaml
in your current working directory. -
Fill in your cloud provider specific information.
- Azure (CLI)
- Azure (Portal)
- GCP (CLI)
- GCP (Console)
You need several resources for the cluster. You can use the following
az
script to create them:RESOURCE_GROUP=constellation # enter name of new resource group for your cluster here
LOCATION=westus # enter location of resources here
SUBSCRIPTION_ID=$(az account show --query id --out tsv)
SERVICE_PRINCIPAL_NAME=constell
az group create --name "${RESOURCE_GROUP}" --location "${LOCATION}"
az group create --name "${RESOURCE_GROUP}-identity" --location "${LOCATION}"
az ad sp create-for-rbac -n "${SERVICE_PRINCIPAL_NAME}" --role Owner --scopes "/subscriptions/${SUBSCRIPTION_ID}/resourceGroups/${RESOURCE_GROUP}" | tee azureServiceAccountKey.json
az identity create -g "${RESOURCE_GROUP}-identity" -n "${SERVICE_PRINCIPAL_NAME}"
identityID=$(az identity show -n "${SERVICE_PRINCIPAL_NAME}" -g "${RESOURCE_GROUP}-identity" --query principalId --out tsv)
az role assignment create --assignee-principal-type ServicePrincipal --assignee-object-id "${identityID}" --role 'Virtual Machine Contributor' --scope "/subscriptions/${SUBSCRIPTION_ID}"
az role assignment create --assignee-principal-type ServicePrincipal --assignee-object-id "${identityID}" --role 'Application Insights Component Contributor' --scope "/subscriptions/${SUBSCRIPTION_ID}"
echo "subscription: ${SUBSCRIPTION_ID}
tenant: $(az account show --query tenantId -o tsv)
location: ${LOCATION}
resourceGroup: ${RESOURCE_GROUP}
userAssignedIdentity: $(az identity show -n "${SERVICE_PRINCIPAL_NAME}" -g "${RESOURCE_GROUP}-identity" --query id --out tsv)
appClientID: $(jq -r '.appId' azureServiceAccountKey.json)
clientSecretValue: $(jq -r '.password' azureServiceAccountKey.json)"Fill the values produced by the script into your configuration file.
By default, Constellation uses
Standard_DC4as_v5
CVMs (4 vCPUs, 16 GB RAM) to create your cluster. Optionally, you can switch to a different VM type by modifying instanceType in the configuration file. For CVMs, any VM type with a minimum of 4 vCPUs from the DCasv5 & DCadsv5 or ECasv5 & ECadsv5 families is supported.Run
constellation config instance-types
to get the list of all supported options.-
subscription: The UUID of your Azure subscription, e.g.,
8b8bd01f-efd9-4113-9bd1-c82137c32da7
.You can view your subscription UUID via
az account show
and read theid
field. For more information refer to Azure's documentation. -
tenant: The UUID of your Azure tenant, e.g.,
3400e5a2-8fe2-492a-886c-38cb66170f25
.You can view your tenant UUID via
az account show
and read thetenant
field. For more information refer to Azure's documentation. -
location: The Azure datacenter location you want to deploy your cluster in, e.g.,
westus
. CVMs are currently only supported in a few regions, check Azure's products available by region. These are:westus
eastus
northeurope
westeurope
-
resourceGroup: Create a new resource group in Azure for your Constellation cluster. Set this configuration field to the name of the created resource group.
-
userAssignedIdentity: Create a new managed identity in Azure. You should create the identity in a different resource group as all resources within the cluster resource group will be deleted on cluster termination.
Add two role assignments to the identity:
Virtual Machine Contributor
andApplication Insights Component Contributor
. Thescope
of both should refer to the previously created cluster resource group.Set the configuration value to the full ID of the created identity, e.g.,
/subscriptions/8b8bd01f-efd9-4113-9bd1-c82137c32da7/resourcegroups/constellation-identity/providers/Microsoft.ManagedIdentity/userAssignedIdentities/constellation-identity
. You can get it by opening theJSON View
from theOverview
section of the identity.The user-assigned identity is used by instances of the cluster to access other cloud resources. For more information about managed identities refer to Azure's documentation.
-
appClientID: Create a new app registration in Azure.
Set
Supported account types
toAccounts in this organizational directory only
and leave theRedirect URI
empty.Set the configuration value to the
Application (client) ID
, e.g.,86ec31dd-532b-4a8c-a055-dd23f25fb12f
.In the cluster resource group, go to
Access Control (IAM)
and set the created app registration asOwner
. -
clientSecretValue: In the previously created app registration, go to
Certificates & secrets
and create a newClient secret
.Set the configuration value to the secret value.
-
instanceType: The VM type you want to use for your Constellation nodes.
For CVMs, any type with a minimum of 4 vCPUs from the DCasv5 & DCadsv5 or ECasv5 & ECadsv5 families is supported. It defaults to
Standard_DC4as_v5
(4 vCPUs, 16 GB RAM).Run
constellation config instance-types
to get the list of all supported options.
You need a service account for the cluster. You can use the following
gcloud
script to create it:SERVICE_ACCOUNT_ID=constell # enter name of service account here
PROJECT_ID= # enter project id here
SERVICE_ACCOUNT_EMAIL=${SERVICE_ACCOUNT_ID}@${PROJECT_ID}.iam.gserviceaccount.com
gcloud iam service-accounts create "${SERVICE_ACCOUNT_ID}" --description="Service account used inside Constellation" --display-name="Constellation service account" --project="${PROJECT_ID}"
gcloud projects add-iam-policy-binding "${PROJECT_ID}" --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" --role='roles/compute.instanceAdmin.v1'
gcloud projects add-iam-policy-binding "${PROJECT_ID}" --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" --role='roles/compute.networkAdmin'
gcloud projects add-iam-policy-binding "${PROJECT_ID}" --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" --role='roles/compute.securityAdmin'
gcloud projects add-iam-policy-binding "${PROJECT_ID}" --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" --role='roles/compute.storageAdmin'
gcloud projects add-iam-policy-binding "${PROJECT_ID}" --member="serviceAccount:${SERVICE_ACCOUNT_EMAIL}" --role='roles/iam.serviceAccountUser'
gcloud iam service-accounts keys create gcpServiceAccountKey.json --iam-account="${SERVICE_ACCOUNT_EMAIL}"
echo "project: ${PROJECT_ID}
serviceAccountKeyPath: $(realpath gcpServiceAccountKey.json)"Fill the values produced by the script into your configuration file.
By default, Constellation uses
n2d-standard-4
VMs (4 vCPUs, 16 GB RAM) to create your cluster. Optionally, you can switch to a different VM type by modifying instanceType in the configuration file. Supported are all machines from the N2D family. Refer to N2D machine series or runconstellation config instance-types
to get the list of all supported options.-
project: The ID of your GCP project, e.g.,
constellation-129857
.You can find it on the welcome screen of your GCP project. For more information refer to Google's documentation.
-
region: The GCP region you want to deploy your cluster in, e.g.,
us-west1
.You can find a list of all regions in Google's documentation.
-
zone: The GCP zone you want to deploy your cluster in, e.g.,
us-west1-a
.You can find a list of all zones in Google's documentation.
-
serviceAccountKeyPath: To configure this, you need to create a GCP service account with the following permissions:
Compute Instance Admin (v1) (roles/compute.instanceAdmin.v1)
Compute Network Admin (roles/compute.networkAdmin)
Compute Security Admin (roles/compute.securityAdmin)
Compute Storage Admin (roles/compute.storageAdmin)
Service Account User (roles/iam.serviceAccountUser)
Afterward, create and download a new JSON key for this service account. Place the downloaded file in your Constellation workspace, and set the config parameter to the filename, e.g.,
constellation-129857-15343dba46cb.json
. -
instanceType: The VM type you want to use for your Constellation nodes.
Supported are all machines from the N2D family with a minimum of 4 vCPUs. It defaults to
n2d-standard-4
(4 vCPUs, 16 GB RAM), but you can use any other VMs from the same family. Refer to N2D machine series or runconstellation config instance-types
to get the list of all supported options.
infoIn case you don't have access to CVMs on Azure, you may use less secure trusted launch VMs instead. For this, set confidentialVM to
false
in the configuration file. -
Download the trusted measurements for your configured image.
constellation config fetch-measurements
For details, see the verification section.
-
Create the cluster with one control-plane node and two worker nodes.
constellation create
uses options set inconstellation-conf.yaml
.tipOn Azure, you may need to wait 15+ minutes at this point for role assignments to propagate.
constellation create --control-plane-nodes 1 --worker-nodes 2 -y
This should give the following output:
$ constellation create ...
Your Constellation cluster was created successfully. -
Initialize the cluster
constellation init
This should give the following output:
$ constellation init
Your Constellation master secret was successfully written to ./constellation-mastersecret.json
Initializing cluster ...
Your Constellation cluster was successfully initialized.
Constellation cluster identifier g6iMP5wRU1b7mpOz2WEISlIYSfdAhB0oNaOg6XEwKFY=
Kubernetes configuration constellation-admin.conf
You can now connect to your cluster by executing:
export KUBECONFIG="$PWD/constellation-admin.conf"The cluster's identifier will be different in your output. Keep
constellation-mastersecret.json
somewhere safe. This will allow you to recover your cluster in case of a disaster.infoDepending on your CSP and region,
constellation init
may take 10+ minutes to complete. -
Configure kubectl
export KUBECONFIG="$PWD/constellation-admin.conf"
Deploy a sample application
-
Deploy the emojivoto app
kubectl apply -k github.com/BuoyantIO/emojivoto/kustomize/deployment
-
Expose the frontend service locally
kubectl wait --for=condition=available --timeout=60s -n emojivoto --all deployments
kubectl -n emojivoto port-forward svc/web-svc 8080:80 &
curl http://localhost:8080
kill %1
Terminate your cluster
constellation terminate
This should give the following output:
$ constellation terminate
Terminating ...
Your Constellation cluster was terminated successfully.
On Azure, if you have used the az
script, you can keep the prerequisite resources and reuse them for a new cluster.
Or you can delete them:
RESOURCE_GROUP=constellation # name of your cluster resource group
APPID=$(jq -r '.appId' azureServiceAccountKey.json)
az ad sp delete --id "${APPID}"
az group delete -g "${RESOURCE_GROUP}-identity" --yes --no-wait
az group delete -g "${RESOURCE_GROUP}" --yes --no-wait