Ask an OpenShift Admin Office Hour - Bare metal deployments

What's the worst part about provisioning bare metal machines? And why would you opt to deploy Red Hat OpenShift on a bare metal server over a virtual machine anyway? In this episode we take an in-depth look at bare metal provisioning with OpenShift Container Platform.

More specifically, Rhys Oxenham and Ben Schmaus, both from the Customer and Field Enablement team, along with Ramon Acedo, OpenShift product manager, join us to examine and demo bare metal with the installer-provisioned infrastructure (IPI) method. This includes a look at the prerequisites, and setting up the environment for success. 

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 Red Hat livestreaming 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 45 recorded stream:

Use this link to jump directly to where we start talking about today’s topic. 

This week’s top of mind topics:

• VMworld 2021 happened the week of October 5-7, you can find all of the OpenShift sessions at this link, including one which I co-presented - MCL3142S.
• If you missed it, the OpenShift product management team delivered the “What’s New” session for OpenShift 4.9 on October 7th. This nearly two hour session covers a huge number of the changes and additions to OpenShift in the soon-to-be-released version 4.9.
• By the time you’re reading this KubeCon 2021 North America will have concluded. If you’re interested in Red Hat’s presence - sessions, booth, pre-event events, and more - please visit the landing page here.
• Our last topic for this week is the poison pill Operator. The docs were recently added for this Operator, so it’s relatively unknown at this point. The short version is that this Operator provides a method to quickly identify when a node has failed and reschedule workload which was hosted on it.

Questions answered and topics discussed during the stream:

• We start this week’s topic by asking Ramon when and where we see customers using bare metal and what use cases we foresee.
• Metal3 provides a machine API interface for provisioning and managing hardware servers with OpenShift. We talk about this during the stream here.
• Managing servers at scale is hard, Metal3 and OpenShift is the way we make that easier.
• One of our viewers asked what are the reasons to use OpenShift instead of “vanilla” Kubernetes? In addition to the answer we provided during the stream, there is an excellent e-book, written by Joe Fernandes, vice president for hybrid platforms, that goes in detail on the differences between OpenShift and Kubernetes, along with why customers choose OpenShift.
• Rhys begins to demo an OpenShift bare metal IPI cluster here during the stream. During the demo, he covers many aspects of using a cluster with the bare metal cloud provider including
  • The bare metal host details , which gives us the ability to see information about the hardware, its status, and do management tasks like rebooting.
  • An explanation of the installation parameters used to connect to the hardware for provisioning.
  • A tour of the Metal3 functions deployed to the cluster and how they manage the hardware.
  • A dive into the Ironic functions used by Metal3, including showing the Ironic interfaces that talk with the hardware.
  • More details about individual servers surfaced up by the cloud provider, such as NICs, hardware status, power status, and more.
  • Adding more nodes to the cluster using the cloud provider is done by creating a new ConfigMap with the machine’s details. Once that’s done, the node is added to the inventory and is ready to be configured and used for workload.
  • Scaling the cluster by adding the available node.
• A viewer asks about CRDs which need to be deployed to utilize the bare metal cloud provider. There’s nothing specific to install since the installer will configure everything when the cluster is deployed. Put differently, when you do a bare metal IPI deployment, all of the components needed for the cloud provider are automatically there.
• Is bare metal IPI related to the Zero Touch Provisioning (ZTP) functionality coming with Advanced Cluster Manager? No, aside from IPI is used by ZTP to deploy clusters.
• What about single node OpenShift (SNO) with bare metal IPI? Single node OpenShift is, as the name implies, a single node. As a result, there would not be a reason to use the cloud provider to scale the cluster up.
• While we demoed using IPMI to manage the hardware, you can also use Redfish. This changes the hardware management slightly, but the functionality is the same - the ability to manage, scale, and utilize hardware with your OpenShift cluster. Redfish also adds the ability to use virtual media for booting and installing the servers, which might be a better option if PXE is not available in your network.