The lifecycle of the Virtual Machines and their relationship to the Kubernetes cluster will be managed using two operators: vNode-Operator(ViNO), and the Support Infra Provider Operator (SIP).
ViNO is responsible for setting up VM infrastructure, such as:
rack: 8
and node: rdm8r008c002
- should follow k8s semi-standardnode-type: worker
node-flavor: foobar
networks: [foo, bar]
and the details for ViNO can be found hereThe Cluster Support Infrastructure Provider, or SIP, is responsible for the lifecycle of:
BareMetalHost
resources to label (or unlabel) based on scheduling
constraints.BareMetalHost
objects to use in the creation of supporting
infrastructure.ViNO is a project written in Go, and the make targets used to deploy ViNO leverage both Go and Kustomize commands which require Golang be installed.
For detailed installation instructions, please see the Golang installation guide.
In order to apply manifests to your cluster via Make targets we suggest the use of Kustomize.
For detailed installation instructions, please see the Kustomize installation guide.
If your organization requires development behind a proxy server, you will need to define the following environment variables with your organization's information:
HTTP_PROXY=http://username:password@host:port
HTTPS_PROXY=http://username:password@host:port
NO_PROXY="localhost,127.0.0.1,10.96.0.0/12"
PROXY=http://username:password@host:port
USE_PROXY=true
10.96.0.0/12 is the Kubernetes service CIDR.
The default vino-builder VM flavors use 1G hugepages. To ensure hugepages support is configured correctly at the host level, do the following:
For Ubuntu 18.04 and earlier, install the 'hugepages' package:
# sudo apt install hugepages
For later ubuntu versions, install the 'libhugetlbfs-bin' package:
# sudo apt install libhugetlbfs-bin
Then run:
# hugeadm --explain
This will display a block like the following describing the current hugepages configuration:
Total System Memory: 96676 MB
Mount Point Options
/dev/hugepages rw,relatime,pagesize=1024M
Huge page pools:
Size Minimum Current Maximum Default
1073741824 64 64 64 *
In the example above, pagesize=1024M indicates a default 1GB page size for the /dev/hugepages mount, and 64 pages are preallocated for 64GB available memory. If the default pagesize is other than 1024M or the number of preallocated pages is insufficient for the subcluster VMs to be created, add the parameters default_hugepagesz, hugepagesz and hugepages to the end of GRUB_CMDLINE_LINUX_DEFAULT in your host's grub config file. For example,
# Set the default commandline
GRUB_CMDLINE_LINUX_DEFAULT="console=tty1 console=ttyS0 default_hugepagesz=1G hugepagesz=1G hugepages=64"
(In a typical Ubuntu installation, the grub configuration can usually be found at /etc/default/grub or somewhere under /etc/grub.d. If the host is a VM built from a cloud image, the grub config may be found at /etc/default/grub.d/50-cloudimg-settings.cfg.)
After making changes, run the following to make them take effect:
sudo update-grub
sudo reboot now
Airship projects often have to deploy Kubernetes, with common requirements such as supporting network policies or working behind corporate proxies. To that end the community maintains a Kubernetes deployment script and is the suggested way of deploying your Kubernetes cluster for development purposes.
# curl -Lo deploy-k8s.sh https://opendev.org/airship/charts/raw/branch/master/tools/gate/deploy-k8s.sh
# chmod +x deploy-k8s.sh
# sudo ./deploy-k8s.sh
When Kubernetes is deployed from the script above it installs Docker but does not configure it to run as a non-root user. The shell commands below are optional to configure Docker to run as a non-root user. They include creating the docker group, adding the current user to that group updating the group without having to log out and testing functionality with the hello-world container.
If you choose to skip these steps, please continue with the developer environment steps as a root user.
# sudo groupadd docker
# sudo usermod -aG docker $USER
Log out and log back in again for the changes to take effect, then test functionality with a hello world container.
# docker run hello-world
Once your cluster is up and running, you'll need to build the ViNO image to use, and to deploy the operator on your cluster:
# make docker-build-controller
# make deploy
Once these steps are completed, you should have a working cluster with ViNO deployed on top of it:
# kubectl get pods --all-namespaces
NAMESPACE NAME READY STATUS RESTARTS AGE
kube-system calico-kube-controllers-7985fc4dd6-6q5l4 1/1 Running 0 3h7m
kube-system calico-node-lqzxp 1/1 Running 0 3h7m
kube-system coredns-f9fd979d6-gbdzl 1/1 Running 0 3h7m
kube-system etcd-ubuntu-virtualbox 1/1 Running 0 3h8m
kube-system kube-apiserver-ubuntu-virtualbox 1/1 Running 0 3h8m
kube-system kube-controller-manager-ubuntu-virtualbox 1/1 Running 0 3h8m
kube-system kube-proxy-ml4gd 1/1 Running 0 3h7m
kube-system kube-scheduler-ubuntu-virtualbox 1/1 Running 0 3h8m
kube-system storage-provisioner 1/1 Running 0 3h8m
vino-system vino-controller-manager-788b994c74-sbf26 2/2 Running 0 25m
To configure the bare metal networking interface that should be used for the VM Bridge, please specify it in your vino CR at field spec.vmBridge.
# kubectl apply -f config/samples/vino_cr.yaml
# kubectl -n vino-system get pods
# kubectl -n vino-system get ds
delete vino CR and make sure DaemonSet is deleted as well
# kubectl delete vino vino-test-cr
# kubectl -n vino-system get ds
# kubectl -n vino-system get cm
For any questions on the ViNo, or other Airship projects, we encourage you to join the community on Slack/IRC or by participating in the mailing list. Please see this Wiki for contact information, and the community meeting schedules.