Support vertical scale up for master nodes on RHOSP IPI

Request:

Same request as RFE-1000 (see full description below) for Red Hat OpenStack Platform.

History:

• Although the status of RFE-1000 was set as rejected, the feature is being implemented on OCPPLAN-5712.
• At some point, OCPPLAN-5712 got focused only on AWS and it was requested to submit new RFEs for other IPI platforms.

Description:

1) Proposed title of this feature request

[RHOCP4] Support vertical scale up for master node

2) What is the nature and description of the request?

Please support vertical scale up(CPU, Memory, disk) for master node after installation.

3) Why does the customer need this? (List the business requirements here)

Our customer wants to start OCP4 with fewer worker nodes, and then add additional worker nodes as needed. The maximum number of worker nodes we can add depends on the size of master node, so to scale up master node would be required in future when adding more additional nodes than they expected at the beginning. To replace master nodes from small to large one by one could be a solution of this, but it's difficult to do on baremetal env. (Even on VM env, it is sometimes difficult because it requires extra resources temporally for +1 node) To do scale up master node vertically('nova resize' command in OpenStack) is simple and easy way for customer. Although each master node will goes down during vertical scale up, that the same as during rolloing upgrade("oc adm upgrade"). So we think there would be no hurdles for Red Hat to support vertical scale up for master node.

4) Do you have any specific timeline dependencies and which release would they like to target?

Our customer would start OCP4 with OCP4.4 or 4.5, so we would like Red Hat to support vertical scale up on that versions.

5) Would you be able to assist in testing this functionality if implemented?

Yes. we can assist.

6) How would you like to achieve this? (List the functional requirements here)?

Red Hat supports vertical scale up(CPU, Memory, disk) for master node officially, and such fact is written in OCP4 manual - User can do vertical scale up master nodes and cluster can run even after the scale up.

7) For each functional requirement listed in question 6, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented.

Check if manual describes that vertical scale up is supported - Check if "oc get nodes", "oc get co" and other oc cmmand can work after the scale up