Ask an OpenShift Admin Office Hour - Installation methods redux

Last year, over the course of several episodes, we discussed many of the different installation platforms and options available for OpenShift. Since then there have been some updates, changes, and newly added capabilities. This episode we are joined by Katherine Dubé, product manager for Red Hat OpenShift focused on the installer, to discuss those changes, highlights some new options, and provides some guidance on choosing an infrastructure and install method for your clusters.

As always, please see the list below for additional links to specific topics, questions, and supporting materials for the episode!

If you’re interested in more streaming content, please subscribe to the OpenShift.tv streaming calendar to see the upcoming episode topics and to receive any schedule changes. If you have questions or topic suggestions for the Ask an OpenShift Admin Office Hour, please contact us via Discord, Twitter, or come join us live, Wednesdays at 11am EDT / 1500 UTC, on YouTube and Twitch.

Episode 25 recorded stream:

Supporting links for today’s topic:

• Use this link to jump straight to where we begin talking about the installer and installation process.
• How to choose a cluster installation type, from the documentation.

Other links and materials referenced during the stream:

• The Red Hat OpenShift Container Platform Architecture Design Guide for PCI-DSS has been updated!
• Katherine mentioned using OpenShift Hive as an API-driven method of deploying and managing OpenShift clusters.
• You can find information about what was added to the most recent OpenShift release, as well as what’s on the roadmap here: https://www.openshift.com/nextandnew 

Questions answered during the stream:

• What tools and documentation is available for migrating from OpenShift 3 to OpenShift 4? We have a topic page specifically for migration here, where you can find lots of additional information too.
• How do I add new nodes to a cluster deployed using the assisted installer? Follow the standard process for deploying nodes with the assisted installer - boot to the ISO. But, you can also add nodes just like with UPI.
• What is the difference between IPI and UPI? Are there advantages or disadvantages to one or the other? And, what, if any, differences are there in the cluster when it’s done being installed?
• IPI allows scaling with a CLI command or using autoscalers, does that make it superior to UPI?
• The debate between which is superior - IPI vs UPI - rages on!
• The scope of the installer changed with OpenShift 4 to focus only on, well, installing the cluster. The other tasks were moved to day 2! Katherine goes in-depth on how and why the install process changed.
• How can I host multiple clusters for remote access, for example, to let my students use and learn OpenShift?
• A question we didn’t get to answer during the stream: can ACM deploy OpenShift clusters in disconnected environments? Yes! ACM supports disconnected mirror registry. You can find the docs for the update/upgrade process here, and for vSphere specifically here.
• Do Windows worker nodes support UPI deployments? Not yet, but it’s on the roadmap!
• Does the installer allow for mixed IPI and UPI deployments? If not, is there an easy way to add manually provisioned nodes to an IPI cluster? The installer doesn’t support mixing the install types, but you can add manually provisioned nodes to an IPI-provisioned cluster at any time - so long as they have the same cloud provider integration. For example, if you deploy vSphere IPI, then you can only add nodes which have been provisioned using the vSphere integrations. You could not add a node that is, for example, a physical server.
• Can you convert between IPI and UPI? Katherine points out that IPI and UPI are just install methods - once the cluster is provisioned, you can manage both deployment types the same!
• Can I deploy IPI with an external load balancer? Unfortunately, no. Not yet in a supported way. But, the assisted installer folks are aware of this request.
• What are some reasons customers push back against, or choose not to use, the IPI install method?
• Can I add nodes from different infrastructure types? Only if all nodes are deployed using the non-integrated, platform agnostic method.
• When deploying using IPI with vSphere, is it possible for OpenShift to configure (anti)affinity rules for control plane nodes? No, unfortunately not today.