As of version 8.15.6, Turbonomic now supports virtual machines deployed in your OpenShift clusters via Red Hat OpenShift Virtualization. One of the new features we support is virtual machine moves, which are non-disruptive. This means your workloads will not see any interruptions when executing these actions (documentation can be found here).

Introduction

Managing resources for your OpenShift cluster is vital, especially for the nodes hosting all your workloads. Fail to manage these resources correctly and you may not be able to execute resize actions on workloads when you need them the most, subsequently leaving applications starved for resources. This is where move actions come in.

OpenShift Virtualization virtual machines in Turbonomic

In Turbonomic, the representation of an OpenShift Virtualization virtual machine is a workload controller. If you look at the supply chain of one of your OpenShift Virtualization virtual machines in Turbonomic, you will see the following: 

Note that workload controller entities have an associated container pod, which is deployed on a virtual machine instance. In our case, the virtual machine entities in Turbonomic are the representations of the nodes in the corresponding OpenShift cluster.

To execute a move on an OpenShift Virtualization virtual machine in Turbonomic, you will need to navigate to the container pod associated with a workload controller. It is these container pods in Turbonomic which will contain the move actions that you can execute.

Now, you might ask, “Why are move actions found on container pods in Turbonomic?”.

The reasoning is simple. In OpenShift, container pods are the smallest compute unit that can be defined, deployed, and managed. When you create virtual machines on your OpenShift cluster, it is these pods which will be run (one per vm) and ultimately placed onto various worker nodes. When there is congestion on a node, it is these container pods which will need to be moved from one worker node to another. 

Move Actions in Turbonomic

Once you have a move action generated for one of your container pods, you will see details resembling those found in the following image.

In these images, there are a few key points to make note of. First are the virtual machine entities involved in the action. The virtual machine on which the container pod resides, in this case, is worker2.ocpv.cp.fyre.ibm.com. The virtual machine which Turbonomic is planning on moving the container pod to is named worker0.ocpv.cp.fyre.ibm.com. The second detail to observe are the cpu and memory impacts from the proposed changes. In this case, the source virtual machine will see a reduction in resource usage, while the destination vm will see a growth in resource usage. 

Finally, you can execute the action. Upon completion you should see your virtual machine running on the new host in OpenShift completely uninterrupted! The changes to the data in Turbonomic, for example the new pod name as well as the updated virtual machine entity, will be updated on the next discovery cycle. Using this feature is as simple as adding an OpenShift target (with OpenShift Virtualization installed) to your Turbonomic instance (refer to documentation here). If you are not already a Turbonomic customer, you could book a live demo or start a free trial by signing up here.