Join us for the second online summit of KubeVirt contributors and users online, Feb 16-17
About this event
KubeVirt Summit 2022 Day 1
The KubeVirt Project is holding its second Online Summit February 16-17, here on Community.CNCF.io. Join us to discuss KubeVirt development, share user experiences, learn about new features, and level up your project contributions.
Join us for this Summit! Please register for both DAY 1 and DAY 2
This is a CNCF Community event.
When
Wednesday, February 16, 2022 2:00 PM – 7:00 PM (UTC)
Agenda
2:00 PM
Introduction: KubeVirt Today
Maintainer Fabian Deutch shares where KubeVirt is today.
2:25 PM
KubeVirt on Arm64 (Progress, Troubles and Plans)
Session Leader: Howard Zhang
Currently we have experimental enablement of kubeVirt on Arm64. I would like to give a review on the enablement progress, what troubles we met and how we solved them. Hopefully, it can help you if you are plan to do some multi-arch works. This talk would include following topics:
3:00 PM
EdgeStack and Singulus: Complete VM network implementation with KubeVirt CNI
Sangho Shin will present:
"EdgeStack is a cloud stack for edge environment, where services are distributed in multiple sites but with smaller scale than central data center. EdgeStack is designed with lightweight cloud architecture, which requires compact control nodes. However, EdgeStack support large scale deployment and carrier grade high availability for enterprise level cloud services. EdgeStack provides VM and Managed Kubernetes to support both VM and container workloads. They are managed by single Kubernetes technology by utilizing KubeVirt and Cluster-API technologies. Furthermore, we are planning to integrate the two to single KubeVirt technology in near future.
Although, kubevirt supports VM provisioning by leveraging containerized hypervisor, the VM connectivity is still provided through the existing CNI implementations. Unfortunately, none of the existing CNIs implement the VM networking model, therefore, well-known networking features like tenant based L2, L3 isolation, security group and IP preservation are not supported. To eliminate those limitation, we introduce singulus network stack which aims to provide VPC like networking model for kubevirt. With singulus network stack, the end-users can seamlessly migrate from OpenStack to kubevirt without compromising any networking features."
3:30 PM
Live Migration Policies - fine tuning migrations dynamically
Itamar Holder leads a discussion of Live Migration Polices, including architecture and goals (50min)
4:35 PM
KubeVirt+Harverster: a Windows user tale
Nuno do Carmo leads a session on:
"Can we run Windows Server on KubeVirt? of course we can, however what about UEFI, GPT partioning scheme, SecureBoot? Since the open source project Harvester hit the v1.0 milestone, this tutorial will guide you on how to create a Virtual Machine, running on KubeVirt, and using advanced, yet simple, configuration to make a Windows VM feeling even more ""Windows"". Another aspect of this tutorial was to share the lessons learnt along the way, as some settings while explained, might be difficult to find sometimes."
5:10 PM
Kubernetes clusters on KubeVirt VMs
David Vossel and co-presenters will share:
The cluster-api-provider-kubevirt is a new community project that aims at extending the Kubernetes Cluster API to provide the ability to spin up Kubernetes clusters on top of KubeVirt VMs. In this presentation I'll discuss the design of the Kubernetes Cluster API and how KubeVirt fits into this picture. To aid in this discussion, I'll demonstrate how to use the Cluster API tooling to declaratively provision new clusters on KubeVirt VMs.
Attendees will walk away from this presentation understanding the basics of how to manage Kubernetes clusters within KubeVirt VMs as well as an understanding of the potential this functionality unlocks.
5:45 PM
Network interface hotplug for KubeVirt
Miguel Duarte Barroso will lead a session on:
"This talk will provide the audience with a basic understanding of KubeVirt, CNI, and Multus, and then propose a design to add (or remove) network interfaces from running pods (and virtual machines), along with the changes required in Multus and KubeVirt to make it happen. It will also factor in a community perspective, explaining how we pitched and got both the Multus and KubeVirt communities involved in a working arrangement to deliver this functionality."
6:20 PM
A few bugs and findings from VMI Churn at NVIDIA
Fan Zhang will lead:
Illustrate NVIDIA’s production GPU Cloud use-case leveraging Kubevirt and how we found some interesting bugs. It includes : VMIs stuck in running status while the virt-launcher pod were gone. virt-handler not able to pick up the correct launcher socket because the ghost record did not clean up properly. Deleting lots of VMIs caused virt-controler panic at a large scale (1000+ VMIs). The audience can learn some stories about how Kubevirt is iterating to adjust to a large scale and rapid, not durable workload.
