Skip to content

Latest commit

 

History

History
 
 

cluster

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
arbiter
arbiter
 
 
nodes
nodes
 
 
README.md
README.md
 
 

README.md

Crafter CMS Authoring Cluster with embedded DB

Enterprise images are required for running an Authoring cluster with embedded DB

Start the cluster

  1. cd nodes
  2. Run kubectl apply -k .
  3. cd ../arbiter
  4. Run kubectl apply -k .

Update the cluster nodes

IMPORTANT: These steps are mainly for making minor configuration updates. If you need to upgrade the Crafter CMS images to a new version, please follow the Upgrade to a new version of Crafter CMS section

  1. cd nodes
  2. Run kubectl apply -k. The containers are not automatically terminated.
  3. Terminate one container so that a new container with the updated deployment changes is created.
  4. Terminate the second container after the first one is ready. A new container is going to be created.

Shutdown the cluster

  1. cd arbiter
  2. Run kubectl delete -k .. Make sure the container has been completely terminated
  3. cd ../nodes
  4. Run kubectl delete -k . --cascade=false
  5. Terminate one of the containers
  6. Make sure the first container has been completely terminated, then terminate the second one

Upgrade to a new version of Crafter CMS

  1. Scale the Arbiter Deployment to 0, by updating the spec.replicas in the arbiter/arbiter-deployment.yaml and runnin kubectl apply -k .. Make sure the arbiter pod is terminated.
  2. Scale the Authoring StatefulSet to 1, by updating the spec.replicas in the nodes/authoring-deployment.yaml, making sure the MARIADB_CLUSTER_NODE_COUNT env variable is set to 1, and running kubectl apply -k .. Verify the authoring-1 (second pod) is terminated. If not, kill it. Only authoring-0 should be running now.
  3. Change the image tags to the new Crafter CMS version in the YAML files.
  4. Upgrade nodes/resources/config/studio configuration files to their new versions.
  5. cd nodes and run kubectl apply -k . (make sure spec.replicas and MARIADB_CLUSTER_NODE_COUNT are still 1, it's important that authoring-1 is NOT started since it's our backup in case the upgrade goes wrong). If the authoring-0 pod is not automatically terminated, terminate it. authoring-0 should be restarted.
  6. Watch the log of authoring_tomcat container of the authoring-0 pod for upgrade manager errors.
  7. If the upgrade manager fails:
    1. Save the /opt/crafter/data, the /opt/crafter/logs and the container log of the authoring_tomcat container. You can use these later for troubleshooting.
    2. Scale the Authoring StatefulSet to 0. No Authoring pod should be running.
    3. Delete the data-authoring-0 pvc (Persistent Volume Claim). authoring-0 will start with no data.
    4. Scale the Authoring StatefulSet to 2. authoring-1 will bootstrap the cluster with the data previous to the upgrade and authoring-0 will replicate it.
  8. If the upgrade manager finished successfully, you can now scale the Authoring StatefulSet back to 2 so authoring-1 replicates the upgrade.
  9. Scale the Arbiter Deployment back to 1.