Skip to content

scale-out-mno.md

Latest commit

 

History

History
49 lines (32 loc) · 3.67 KB

scale-out-mno.md

File metadata and controls

49 lines (32 loc) · 3.67 KB

Scale Out a Multi-Node Openshift Deployment

A JetLag deployed Multi-Node Openshift deployment can be scale out via JetLag. Workers can be added using JetLag Inventory and Playbooks. This guide assumes you have an existing OCP cluster deployed via JetLag. The worker section in the JetLag inventory file should contain records that represent the worker nodes currently joined to the running cluster.

Steps to Scale Out:

Add Nodes to Worker Inventory

To add new node entries to the worker inventory there are three potential options.

  1. New bare metal nodes added to cloud assignment

    If more nodes were added to your lab assigment, update worker_node_count in the ansible/vars/all.yml file and rerun the create-inventory playbook. Be sure to compare the previous inventory file to the new one to ensure that everything is the same except the new nodes added to the worker section.

  2. New virtual nodes from hv_vm inventory

    If you have enabled and setup hv_inventory nodes/and VMs you can use these VMs as new nodes to scale out workers. Records to represent the VMs you would like to scale to must be added to the end of the worker section. If you have not used any of the VMs for anything other than scale out you can simply increase the hybrid_worker_count value in ansible/vars/all.yml and rerun the create-inventory playbook. Be sure that the appropriate playbooks have been run to ensure the VM entries being added to the worker inventory section have been created on the hv nodes. Careful, the playbooks to manupulate VMs operate on all hv_vm records at one time. Be sure to compare the previous inventory file to the new one to ensure that everything is the same except the new nodes added to the worker section.

  3. Manual Entry

    You can add new entries to the worker inventory section manually. Place them at the end of the list of worker entries.

The new nodes, baremetal or virtual, must be placed at the end of the worker nodes inventory. The scale out playbook is designed to use the last n nodes in the inventory.

Update scale_out.yml

There are two variables in ansible/vars/scale_out.yml that indicate which entries from the worker inventory section should be added to the existing cluster.

  • current_worker_count: This value indicates the number of entries in the worker inventory section to skip before starting to add nodes to the existing cluster. This number should match the current number of worker nodes associated with the existing cluster.
  • scale_out_count: This value indicates the number of entries in the worker inventory section that will be added as new workers to the existing cluster.

Example: If the initial OCP deployment had three baremetal workers and the intended worker count was ten, current_worker_count would be 3 and scale_out_count would be 7. Scale out from three existing workers, adding seven new workers, for a total of ten worker nodes.

Run ocp-scale-out.yml

Once the new worker records are added to the inventory and the scale_out.yml file has the proper values. The final step is to run the mno-scale-out.yml playbook.

(.ansible) [root@xxx-h01-000-r650 jetlag]# ansible-playbook -i ansible/inventory/cloud99.local ansible/ocp-scale-out.yml
...

This playbook will:

  • Generate node configuration yml
  • Invoke oc adm node-image create with the node configuration, which generates a discovery ISO
  • Boot the new worker nodes off of the generated discovery ISO
  • Approved generated CSRs

This workflow can be run repeatedly to add more workers to the existing cluster.