kubernetes

DEPRECATED

This page is now deprecated.
Kubernetes deployments of Artifactory and other JFrog products can be done with our official Helm Charts.

If you want to avoid using Helm and use only bare Kubernetes templates, you can use helm template to generate a full template from the chart and deploy with it

# Get the chart
helm fetch jfrog/artifactory

# Generate the Kubernetes template
helm template artifactory-VERSION.tgz > artifactory-template.yaml

# Deploy using the template
kubectl apply -f artifactory-template.yaml

Read the chart's documentation on ways to customise the templates.

---

Artifactory in Kubernetes examples

This directory has some examples for setting up Artifactory running in a Kubernetes cluster.

Kubernetes

Kubernetes is an open-source system for orchestrating containerized applications. To learn more about Kubernetes, see details in the Kubernetes documentation.
This page assumes you have prior knowledge of Kubernetes and have a working cluster to deploy in.

Helm - a package manager for Kubernetes

The recommended way to deploy your applications to Kubernetes is using Helm charts (packages).
Artifactory Pro can be deployed and managed by the Helm package manager (also supported by Artifactory)
See the helm/artifactory directory for an example and usage.

Kubectl

The examples here are defined and deployed using the kubectl command line tool. See more details in the kubectl documentation.
Also see a useful cheat sheet with a good summary of the useful commands and usage.

In these examples Kubernetes objects are defined as Yaml files, so applying them is a simple call to kubectl apply or kubectl create.

---

Persistent Storage

For persistent storage, all volumes are mounted from the cluster's hosts or as Google disks.
NOTE: The examples here use a simple PersistentVolume and PersistentVolumeClaim for example purposes. This setup should NOT be used for production! You should find your best matching storage solution and use it.

---

Database Driver

The databases used in these examples are PostgreSQL and MySQL.
For Artifactory to communicate with the database, it needs the database driver in its Tomcat's lib directory.

Artifactory Docker image comes with the PostgreSQL driver pre-loaded.

For MySQL, you should build your own Artifactory Docker image using the Dockerfile.mysql in this directory that already adds the driver.
To build the image

# For MySQL
$ docker build -t ${YOUR_DOCKER_REGISTRY}/jfrog/artifactory-pro-mysql:${VERSION} -f Dockerfile.mysql .

This will build an image of artifactory-pro that includes the MySQL driver in it. Make sure to push it into your registry

# For MySQL
$ docker push ${YOUR_DOCKER_REGISTRY}/jfrog/artifactory-pro-mysql:${VERSION}

And edit the artifactory.yml to use this image.

Using Different Databases

Artifactory can run with other databases. For more details on supported databases and how to set them up for use with Artifactory, please refer to Changing the Database in the JFrog Artifactory Use Guide.

---

Deploying your Artifactory to Kubernetes

The following describes the steps to do the actual deployment of the Artifactory and its services to Kubernetes.

Memory and CPU resources

To have full control of the memory and cpu allocated to your applications, it is recommended to set resource requests and limits to all your pods.

All templates here include examples of such resources definitions. The provided are examples. You should tune them to your actual needs.

Preparing other Resources

Need to create some Kubernetes resources that will be used by Nginx as SSL and Artifactory reverse proxy configuration

Docker registry secret

In case you built your own Artifactory image and pushed it to your private registry as suggested above, you might need to define a docker-registry secret to be used by Kubernetes to pull images

$ kubectl create secret docker-registry docker-reg-secret --docker-server=${YOUR_DOCKER_REGISTRY} --docker-username=${USER} --docker-password=${PASSWORD} [email protected]

SSL secret

Create the SSL secret that will be used by the Nginx pod
NOTE: These are self signed key and certificate for demo use only!

$ kubectl create secret tls art-tls --cert=../files/nginx/ssl/demo.pem --key=../files/nginx/ssl/demo.key

You can replace the key and certificate with your own files

$ kubectl create secret tls art-tls --cert=${PATH_TO_CERT}/myssl.pem --key=${PATH_TO_CERT}/myssl.key

Deploying the applications

Now you are ready to create the applications in Kubernetes.
The following sequence deploys

• PostgreSQL or MySQL database
• Artifactory Pro or HA
• Nginx

Note that the resources to use are already defined in the Yaml files.

NOTE: If running on Minikube, you will need to deploy a simpler service (NodePort). See the differences in the code examples below.

Artifactory Pro

Database (using PostgreSQL)

# PostgreSQL storage, pods and service
$ kubectl apply -f postgresql-storage.yml
$ kubectl apply -f postgresql.yml

Artifactory

# Artifactory storage, pods and service
$ kubectl apply -f artifactory-storage.yml
$ kubectl apply -f artifactory.yml

Nginx

# Nginx storage and deployment
$ kubectl apply -f nginx-storage.yml
$ kubectl apply -f nginx-deployment.yml

# Nginx service
# If running on a standard Kubernetes cluster
$ kubectl apply -f nginx-service.yml

# If running on Minikube
$ kubectl apply -f nginx-service-minikube.yml

Once done, you should be able to see the deployed pods and services

# Get pods and their status
$ kubectl get pods
NAME                                          READY     STATUS    RESTARTS   AGE
artifactory-k8s-deployment-1732455857-7w6gw   1/1       Running   0          31m
nginx-k8s-deployment-3171003233-q8gb2         1/1       Running   0          25m
postgresql-k8s-deployment-1240329637-25325    1/1       Running   0          33m

# Get services
# On a standard Kubernetes cluster
$ kubectl get services
NAME                     CLUSTER-IP     EXTERNAL-IP   PORT(S)          AGE
artifactory              10.0.160.189   <nodes>       8081/TCP         31m
kubernetes               10.0.0.1       <none>        443/TCP          3d
nginx-k8s-service        10.0.26.194    59.156.13.6   80/TCP,443/TCP   25m
postgresql-k8s-service   10.0.172.76    <none>        5432/TCP         33m

# On Minikube
$ kubectl get services
NAME                     CLUSTER-IP   EXTERNAL-IP   PORT(S)                      AGE
artifactory              10.0.0.210   <nodes>       8081:32355/TCP               57m
kubernetes               10.0.0.1     <none>        443/TCP                      1h
nginx-k8s-service        10.0.0.113   <nodes>       80:30002/TCP,443:32600/TCP   48m
postgresql-k8s-service   10.0.0.165   <none>        5432/TCP                     1h

Accessing your Artifactory Pro

See Accessing your Artifactory

---

Artifactory HA

Database (using MySQL)

$ kubectl apply -f mysql-storage.yml
$ kubectl apply -f mysql.yml

Artifactory storage

Prepare the storage volumes. One for each node.

$ kubectl apply -f artifactory-ha-storage.yml

Prepare the binary storage configuration

Artifactory HA can be configured with various storage solutions.
You can see more details in Configuring the Filestore.
In thie examples, we deploy a ConfigMap with a simple file-system replication configuration (a cache-fs template).

$ kubectl apply -f artifactory-binarystore.yml

Artifactory Master Key

As of Artifactory 5.7.X and up, the joining of a node to an HA cluster is much simpler. All nodes need to share the same Master Key and database configuration. Create the key in the following way:

$ openssl rand -hex 32

You should put the resulting value inside the files artifactory-ha-node1.yml and artifactory-ha-node2.yml as the value of ARTIFACTORY_MASTER_KEY. The files currently have a default value set, but you should update them for a production deployment.

Artifactory HA nodes

Spin up the two nodes and the Kubernetes service

$ kubectl apply -f artifactory-ha-node1.yml
$ kubectl apply -f artifactory-ha-node2.yml
$ kubectl apply -f artifactory-ha-service.yml

Complete the Artifactory HA cluster setup

Once the nodes are running, you need to complete the setup by installing the licenses for the nodes. You can see more details in Artifactory HA setup.

Check that the primary node (artifactory-node1) is up and ready to work.

# Get the primary node pod name
$ ART_NODE1_POD_NAME=$(kubectl get pods | grep artifactory-ha-node1 | cut -d' ' -f1)

# Follow the log for artifactory-node1
$ kubectl logs -f ${ART_NODE1_POD_NAME}

# Wait for the following to appear in the log:
###########################################################
### Artifactory successfully started (23.275 seconds)   ###
###########################################################

Nginx

Setup the Nginx that is used for load balancing, reverse proxy and SSL handling.

NOTE: Make sure to have the SSL secret create as shown before

# Storage and deployment
$ kubectl apply -f nginx-storage.yml
$ kubectl apply -f nginx-deployment.yml

# Service
# If running on a standard Kubernetes cluster
$ kubectl apply -f nginx-service.yml

# If running on Minikube
$ kubectl apply -f nginx-service-minikube.yml

Once done, you should be able to see the deployed pods and services

# Get pods and their status (example output)
$ kubectl get pods
NAME                                    READY     STATUS    RESTARTS   AGE
artifactory-ha-node1-3776668781-fc3wq   1/1       Running   0          7m
artifactory-ha-node2-3265495874-cz6rj   1/1       Running   0          7m
mysql-k8s-deployment-4196928137-3s6tn   1/1       Running   0          7m
nginx-k8s-deployment-1544469967-bt5m1   1/1       Running   0          1m

# Get services (example output)
# On a standard Kubernetes cluster
$ kubectl get services
NAME                CLUSTER-IP   EXTERNAL-IP   PORT(S)                      AGE
artifactory-node1   10.0.0.159   <nodes>       8081:30690/TCP               8m
artifactory-node2   10.0.0.176   <nodes>       8081:30981/TCP               7m
kubernetes          10.0.0.1     <none>        443/TCP                      3d
mysql-k8s-service   10.0.0.17    <none>        3306/TCP                     8m
nginx-k8s-service   10.0.0.75    59.156.13.6   80:32094/TCP,443:30063/TCP   2m

# On Minikube
$ kubectl get services
NAME                CLUSTER-IP   EXTERNAL-IP   PORT(S)                      AGE
artifactory-node1   10.0.0.159   <nodes>       8081:30690/TCP               8m
artifactory-node2   10.0.0.176   <nodes>       8081:30981/TCP               7m
kubernetes          10.0.0.1     <none>        443/TCP                      3d
mysql-k8s-service   10.0.0.17    <none>        3306/TCP                     8m
nginx-k8s-service   10.0.0.75    <nodes>       80:30002/TCP,443:32600/TCP   2m

Adding more nodes

Adding more Artifactory nodes to your cluster is simple. Here is an example for adding node3

• Add another PersistentVolumeClaim section in artifactory-ha-storage.yml. Give it a new name: artifactory-node3-claim
• Copy artifactory-ha-node2.yml to artifactory-ha-node3.yml. Edit it and rename all node2 to node3
• Deploy the new storage and node

$ kubectl apply -f artifactory-ha-storage.yml
$ kubectl apply -f artifactory-ha-node3.yml

Make sure you have the license needed to have the new node join and activated in the cluster.

---

Accessing your Artifactory

Depending on your deployment type, you can now access Artifactory through its Nginx.

Standard Kubernetes

You can see the Nginx is exposed with a public IP of 59.156.13.6 on ports 80 and 443.
Now just point your browser to http://59.156.13.6/artifactory/ or https://59.156.13.6/artifactory/

Minikube

You need to use the Minikube's IP with the assigned port like 192.168.99.100. You can get the Minikube IP with the command minikube ip.
The assigned minikube ports can be seen in the output of kubectl get services as seen above (port 30002 -> 80 and port 32600 -> 443).

Now point your browser to http://192.168.99.100:30002/artifactory/ or https://192.168.99.100:32600/artifactory/

NOTE: When using https, you might need to confirm trusting the certificate and that will redirect you back to https://192.168.99.100/artifactory, resulting in an error. Just put the port 32600 again in the URL, refresh your page, and Artifactory should now load properly.

NOTE: We have moved our Helm charts to jfrog/charts.