In a recent project I was asked to configure Kubernetes so that it could create and manage DNS records automatically. This automatic management comes via Ingress objects created in Kubernetes. Hostnames specified in ingress objects automatically get DNS records created. This is exactly what the Kubernete’s project ExternalDNS does. Unfortuately, even though there is a section for configuring it on Azure, it wasn’t straightforward to get working. This post walks through getting it up and running and also demonstrates its usage in Azure Government.
How can we modify DNS records for hosts defined in Ingress objects? For intance if we have the following Ingress defined, we want to have Kubernetes add a DNS record for nginx.designingdevops.com. We also want certificates for these new domains to be issued so that web browsers can connect without exception.
We’ll use a number of different Kubernetes projects to implement a solution to this problem. I decided to break the solution up into several different sections with scripts for each section so that it is a little easier to digest. In order to implement this solution we use the following projects:
- NGINX Ingress: Manages incoming requests to Kubernetes
- ExternalDNS: Manages DNS entries in our Azure DNS Zone
- Cert-Manager: Manages Let’s Encrypt certificates for our hosts
At this point we’ll need the Azure CLI installed and a Kubernetes cluster. I have created a public gist that contains all the files I used in this post to create Kubernetes and install ExternalDNS but I leave installing the Azure CLI up to the reader. The gist for my files is available here. To create a Kubernetes cluster you can use the instructions here or the script below.
You will notice in the script above that all the variables are configured with defaults but can be overridden by setting environment variables before running the script. For instance if you are in Azure Commercial the above script will fail since usgovvirginia is the default. If you execute:
>> export LOCATION=eastus >> ./k8s.sh
This will create the Kubernetes cluster in eastus instead of usgovvirginia.
Once we have our cluster we can start installing the components we’ll need in order to manage DNS from Kubernetes:
- NGINX ingress controller
- Azure DNS Zone
- Service Principal with permission to manage DNS Zone
- Azure Public IP we can bind to an ingress controller
- external-dns installed and running
- cert-manager installed and running
Ingress Controller and Public IP
At this point we’ll create an Azure Public IP and assign that IP to a newly installed NGINX ingress controller. I used a static Public IP rather than allow Azure to generate one just so if I decide to rebuild my cluster in the future I can just reassign the same Public IP to the LoadBalancer and not have to worry about DNS record updates.
A careful reader of the script might notice that we use:
- –set controller.publishService.enabled=true and
- –set controller.publishService.pathOverride=ingress/nginx-ingress-controller
These options are used by external-dns so that the correct IP address is assigned to the DNS records. These options are required for external-dns to work correctly.
Now that we have our Ingress controller setup we’ll turn out attention to getting external-dns deployed and configured. You will need to generate a Service Principal and provide it to helm during the installation process. This script will create/return the ids of a resource group and DNS zone. It also provisions a service principal assigns ‘Reader’ permission on the RG and ‘Contributor’ permission on the DNS Zone for the service principal. It then deploys external-dns using the newly create service principal and DNS information.
If you haven’t done so already you should also make sure that your domain has the correct Azure nameservers assigned as your domain’s authoritative nameservers. This is done outside of Azure at your domain registrar.
Last but not least we’ll install and configure cert-manager to manage Let’s Encrypt certificates for us on the hostnames we create in our domain.
Test It Out
If everything has gone according to plan you should now have a Kubernetes cluster that will generate DNS records based on hosts you define in Ingress objects. It will also create SSL certificates for those hosts so that https works correctly and doesn’t generate any host errors.
To test our out new cool deploy you can should create a new file called nginx.yaml that contains the following yaml modified with your own domain name in the Ingress object:
Once you’ve created the nginx.yaml file you can apply it to your Kubernetes cluster. The following command will install the NGINX web service into your current namespace.
>> kubectl apply -f nginx.yaml deployment.apps/nginx created service/nginx-svc created ingress.networking.k8s.io/nginx created
This doesn’t happen right away because it takes time to propogate DNS records but in my experience you should have this resolve correctly within about 30 minutes. Open a web browser and navigate to nginx.yourdomainname.com. You should see the NGINX welcome page als In the meantime you can run the following command to ensure that DNS records have been created for your hostname.
>> az network dns record-set a list -g $DNS_RG -z $DOMAIN_NAME Name ResourceGroup Ttl Type Metadata ------ --------------- ----- ------ ---------- nginx azuredns 300 A
NOTE: Make sure you set the environment variables DNS_RG and DOMAIN_NAME before executing the command above.