Skip to content

Latest commit

 

History

History
124 lines (84 loc) · 8.16 KB

README.md

File metadata and controls

124 lines (84 loc) · 8.16 KB

Eclipse Ditto deployment on Microsoft Azure

This guide describes the most simplistic installation of Eclipse Ditto on Microsoft Azure. It is not meant for productive use but rather for evaluation as well as demonstration purposes or as a baseline to evolve a production grade Application architecture out of it which includes Eclipse Ditto.

Prerequisites

HowTo install Ditto

First we are going to setup the basic Kubernetes infrastructure.

As described here we will create an explicit service principal first. You can add roles to this principal later, e.g. to access a Azure Container Registry (ACR).

service_principal=`az ad sp create-for-rbac --name http://dittoServicePrincipal --skip-assignment --output tsv`
app_id_principal=`echo $service_principal|cut -f1 -d ' '`
password_principal=`echo $service_principal|cut -f4 -d ' '`
object_id_principal=`az ad sp show --id $app_id_principal --query objectId --output tsv`

Note: it might take a few seconds until the principal is available to the cluster in the later steps. So maybe time to get up and stretch a bit.

resourcegroup_name=armtest
az group create --name $resourcegroup_name --location "westeurope"

With the next command we will use the provided Azure Resource Manager (ARM) templates to setup the AKS cluster. This might take a while. So maybe time to try out this meditation thing 😄

Note: Add cosmosDB=true to the az group deployment create call in case you want to opt for the Azure Cosmos DB persistence (see below). AKS cluster name, IP address name for the load balancer as well as virtual network name can be provided as parameter to the template as well.

unique_solution_prefix=myprefix
az group deployment create --name DittoBasicInfrastructure --resource-group $resourcegroup_name --template-file arm/dittoInfrastructureDeployment.json --parameters uniqueSolutionPrefix=$unique_solution_prefix servicePrincipalObjectId=$object_id_principal servicePrincipalClientId=$app_id_principal servicePrincipalClientSecret=$password_principal

Now retrieve secrets and configs from the deployment:

aks_cluster_name=`az group deployment show --name DittoBasicInfrastructure --resource-group $resourcegroup_name --query properties.outputs.aksClusterName.value -o tsv`
ip_address=`az group deployment show --name DittoBasicInfrastructure --resource-group $resourcegroup_name --query properties.outputs.publicIPAddress.value -o tsv`
cosmos_mongodb_primary_master_key=`az group deployment show --name DittoBasicInfrastructure --resource-group $resourcegroup_name --query properties.outputs.cosmosPrimaryMasterKey.value -o tsv`
cosmos_account_name=`az group deployment show --name DittoBasicInfrastructure --resource-group $resourcegroup_name --query properties.outputs.cosmosAccountName.value -o tsv`
public_fqdn=`az group deployment show --name DittoBasicInfrastructure --resource-group $resourcegroup_name --query properties.outputs.publicIPFQDN.value -o tsv`

Now you can set your cluster in kubectl.

az aks get-credentials --resource-group $resourcegroup_name --name $aks_cluster_name

Next deploy helm on your cluster. It will take a moment until tiller is booted up. So maybe time again to get up and stretch a bit.

kubectl apply -f helm-rbac.yaml
helm init --service-account tiller

(Optional): in case you want to use MongoDB with persistent storage as part of your setup we will need a k8s StorageClass with Retain policy.

kubectl apply -f managed-premium-retain.yaml

Next we prepare the k8s environment and our chart for deployment.

k8s_namespace=dittons
kubectl create namespace $k8s_namespace

...and download charts Ditto depends on:

helm dependency update ../helm/eclipse-ditto/

Now install Ditto with helm either with MongoDB as part of the deployment or Azure Cosmos DB as persistence option.

Option 1: Embedded MongoDB

...either with persistent storage as part of the helm release (will be deleted with helm delete):

helm upgrade ditto ../helm/eclipse-ditto/ --namespace $k8s_namespace --set service.type=LoadBalancer,service.loadBalancerIP.enabled=true,service.loadBalancerIP.address=$ip_address,service.annotations."service\.beta\.kubernetes\.io/azure-load-balancer-resource-group"=$resourcegroup_name,mongodb.persistence.enabled=true,mongodb.persistence.storageClass=managed-premium-retain --wait --install

...or with a custom K8s PersistentVolumeClaim independent of the Helm release to ensure the data survives a helm delete:

echo "  storageClassName: managed-premium-retain" >> ../helm/ditto-mongodb-pvc.yaml
kubectl apply -f ../helm/ditto-mongodb-pvc.yaml --namespace $k8s_namespace
helm upgrade ditto ../helm/eclipse-ditto/ --namespace $k8s_namespace --set service.type=LoadBalancer,service.loadBalancerIP.enabled=true,service.loadBalancerIP.address=$ip_address,service.annotations."service\.beta\.kubernetes\.io/azure-load-balancer-resource-group"=$resourcegroup_name,mongodb.persistence.enabled=true,mongodb.persistence.existingClaim=ditto-mongodb-pvc --wait --install

...or without persistence for the MongoDB at all:

helm upgrade ditto ../helm/eclipse-ditto/ --namespace $k8s_namespace --set service.type=LoadBalancer,service.loadBalancerIP.enabled=true,service.loadBalancerIP.address=$ip_address,service.annotations."service\.beta\.kubernetes\.io/azure-load-balancer-resource-group"=$resourcegroup_name --wait --install

Option 2: Azure Cosmos DB's API for MongoDB

Disclaimer: as of now Cosmos DB's API for MongoDB is supported by Ditto only for persistence cases. Ditto's search feature/service does not!

Note: the provided template did create an individual database per service (if you added cosmosDB=true). By default the template provides 400 Request Units (RU) for all collections of Policies, Things and Connectivity and 800 for the Things journal collection. You can override by parameter as well as update RUs later at runtime.

Now install Ditto with Cosmos DB as persistence and Ditto search disabled.

helm upgrade ditto ../helm/eclipse-ditto/ --namespace $k8s_namespace --set search.enabled=false,mongodb.embedded.enabled=false,mongodb.apps.connectivity.uri=mongodb://$cosmos_account_name:$cosmos_mongodb_primary_master_key@$cosmos_account_name.documents.azure.com:10255/connectivity\?ssl=true\&replicaSet=globaldb\&maxIdleTimeMS=120000,mongodb.apps.connectivity.ssl=true,mongodb.apps.things.uri=mongodb://$cosmos_account_name:$cosmos_mongodb_primary_master_key@$cosmos_account_name.documents.azure.com:10255/things\?ssl=true\&replicaSet=globaldb\&maxIdleTimeMS=120000,mongodb.apps.things.ssl=true,mongodb.apps.policies.uri=mongodb://$cosmos_account_name:$cosmos_mongodb_primary_master_key@$cosmos_account_name.documents.azure.com:10255/policies\?ssl=true\&replicaSet=globaldb\&maxIdleTimeMS=120000,mongodb.apps.policies.ssl=true,service.type=LoadBalancer,service.loadBalancerIP.enabled=true,service.loadBalancerIP.address=$ip_address,service.annotations."service\.beta\.kubernetes\.io/azure-load-balancer-resource-group"=$resourcegroup_name --wait --install --timeout 600

Have fun with Eclipse Ditto on Microsoft Azure!

A quick first test (which will return "down" in CosmosDB case as search does not support it):

curl -u ditto:ditto http://$public_fqdn:8080/health