MicroShift with KubeVirt and Kata Containers on Raspberry Pi 4 with AlmaLinux 9 (Lime Lynx)
Introduction
MicroShift is a Red Hat-led open-source community project that is exploring how OpenShift OKD Kubernetes distribution can be optimized for small form factor devices and edge computing. Red Hat Device Edge delivers an enterprise-ready and supported distribution of MicroShift. Red Hat Device Edge is planned as a developer preview early next year and expected to be generally available with full support later in 2023.
Over the last 25 parts, we have worked with MicroShift on multiple distros of Linux on the Raspberry Pi 4 and Jetson Nano. In Part 17, we worked with MicroShift on AlmaLinux 8.5. In this Part 26, we will work with MicroShift on AlmaLinux 9. We will run an object detection sample and send messages to Node Red installed on MicroShift. Further, we will setup KubeVirt and the OKD Web Console and run Virtual Machine Instances in MicroShift. We will also create CentOS 9 Stream and Ubuntu Virtual Machines and use the Containerized Data Importer (CDI). We will run a MongoDB sample and finally finish by building and running Kata containers on MicroShift.
AlmaLinux is a free and open-source Linux distribution, created originally by CloudLinux to provide a community-supported, production-grade enterprise operating system that is binary-compatible with Red Hat Enterprise Linux. Both AlmaLinux and Rocky Linux emerged in response to Red Hat’s December 8, 2020 announcement stating that it will discontinue CentOS based on RedHat releases, instead shifting focus to CentOS Stream, which tracks just ahead of a current RHEL release.
Setting up the Raspberry Pi 4 with AlmaLinux 9
Run the following steps to download the Alma Linux image and setup the Raspberry Pi 4.
- Download the latest AlmaLinux 9 64-bit Arm (aarch64) for use with RPi 4 from http://repo.almalinux.org/rpi/9/images/
- Write to Microsdxc card using balenaEtcher or the Raspberry Pi Imager
- Have a Keyboard and Monitor connected to the Raspberry Pi 4
- Insert Microsdxc into Raspberry Pi4 and poweron
- Login using root/almalinux using the keyboard connected to the Raspberry Pi 4
- Edit the /etc/ssh/sshd_config. Add the following line and restart the ssh daemon
PermitRootLogin yes
systemctl restart sshd
- Find the ethernet dhcp ipaddress of your Raspberry Pi 4 by running the nmap on your Macbook with your subnet
$ sudo nmap -sn 192.168.1.0/24
Nmap scan report for 192.168.1.209
Host is up (0.010s latency).
MAC Address: E4:5F:01:2E:D8:95 (Raspberry Pi Trading)
- Now you can login from your Laptop using ssh with root/almalinux to ipaddress above. You may add your public ssh key to login without password.
$ ssh root@$ipaddress
[root@rpi ~]# mkdir ~/.ssh
[root@rpi ~]# vi ~/.ssh/authorized_keys
[root@rpi ~]# chmod 700 ~/.ssh
[root@rpi ~]# chmod 600 ~/.ssh/authorized_keys
[root@rpi ~]# #chcon -R -v system_u:object_r:usr_t:s0 ~/.ssh/
Check that your key is RSA 2048 or larger with the following command. The RSA 1024 will not work.
ssh-keygen -l -v -f ~/.ssh/id_rsa.pub
If it is 1024, you will get the error
[root@rpi ~]# cat /var/log/secure
Nov 26 16:26:42 rpi sshd[3927]: refusing RSA key: Invalid key length [preauth]
- Extend the partition to maximize disk usage
rootfs-expand
- Update and set the hostname with a domain and add the ipv4 address to /etc/hosts
hostnamectl set-hostname microshift.example.com
echo "$ipaddress microshift microshift.example.com" >> /etc/hosts
dnf update -y
- Optionally, enable wifi
nmcli device wifi list # Note your ssid
nmcli device wifi connect $ssid --ask
- Check the release and file system type.
cat /etc/os-release
lsblk -f
Output:
[root@microshift ~]# cat /etc/os-release
NAME="AlmaLinux"
VERSION="9.1 (Lime Lynx)"
ID="almalinux"
ID_LIKE="rhel centos fedora"
VERSION_ID="9.1"
PLATFORM_ID="platform:el9"
PRETTY_NAME="AlmaLinux 9.1 (Lime Lynx)"
ANSI_COLOR="0;34"
LOGO="fedora-logo-icon"
CPE_NAME="cpe:/o:almalinux:almalinux:9::baseos"
HOME_URL="https://almalinux.org/"
DOCUMENTATION_URL="https://wiki.almalinux.org/"
BUG_REPORT_URL="https://bugs.almalinux.org/"
ALMALINUX_MANTISBT_PROJECT="AlmaLinux-9"
ALMALINUX_MANTISBT_PROJECT_VERSION="9.1"
REDHAT_SUPPORT_PRODUCT="AlmaLinux"
REDHAT_SUPPORT_PRODUCT_VERSION="9.1"
[root@microshift ~]# lsblk -f
NAME FSTYPE FSVER LABEL UUID FSAVAIL FSUSE% MOUNTPOINTS
mmcblk0
├─mmcblk0p1 vfat FAT16 7072-6317 192.7M 33% /boot
├─mmcblk0p2 swap 1 _swap 09f0c50e-3dd9-43d7-80c8-18d379a9fccd [SWAP]
└─mmcblk0p3 ext4 1.0 _/ 5d5b6a44-2817-44e8-910c-5e8aea35d03f 29G 50% /var/lib/containers/storage/overlay
/
- Update the kernel parameters - Concatenate the following onto the end of the existing line (do not add a new line) in /boot/cmdline.txt
cgroup_enable=cpuset cgroup_memory=1 cgroup_enable=memory
A control group (cgroup) is a Linux kernel feature that limits, accounts for, and isolates the resource usage (CPU, memory, disk I/O, network, and so on) of a collection of processes. Cgroups are a key component of containers because there are often multiple processes running in a container that you need to control together. In MicroShift, cgroups are used to implement resource requests and limits and corresponding QoS classes at the pod level.
reboot
Verify
ssh root@$ipaddress
cat /proc/cmdline
mount | grep cgroup # Check cgroup2
cat /proc/cgroups | column -t # Check that memory and cpuset are present
Note that this does not have the hugetlb:
[root@microshift ~]# cat /proc/cmdline
coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=1920 bcm2708_fb.fbheight=1200 bcm2708_fb.fbswap=1 smsc95xx.macaddr=E4:5F:01:2E:D8:95 vc_mem.mem_base=0x3ec00000 vc_mem.mem_size=0x40000000 console=ttyAMA0,115200 console=tty1 root=PARTUUID=5ea2038d-03 rootfstype=ext4 elevator=deadline rootwait cgroup_enable=cpuset cgroup_memory=1 cgroup_enable=memory
[root@microshift ~]# mount | grep cgroup # Check cgroup2
cgroup2 on /sys/fs/cgroup type cgroup2 (rw,nosuid,nodev,noexec,relatime,seclabel,nsdelegate,memory_recursiveprot)
[root@microshift ~]# cat /proc/cgroups | column -t # Check that memory and cpuset are present - Note that this does not have hugepages this will cause errors in microshift logs
#subsys_name hierarchy num_cgroups enabled
cpuset 0 58 1
cpu 0 58 1
cpuacct 0 58 1
blkio 0 58 1
memory 0 58 1
devices 0 58 1
freezer 0 58 1
net_cls 0 58 1
perf_event 0 58 1
net_prio 0 58 1
pids 0 58 1
Install sense_hat and RTIMULib on AlmaLinux
The Sense HAT is an add-on board for the Raspberry Pi. The Sense HAT has an 8 × 8 RGB LED matrix, a five – button joystick and includes the following sensors: Inertial Measurement Unit (Accelerometer, Gyroscope, Magnetometer), Temperature, Barometric pressure, Humidity. If you have the Sense HAT attached, install the libraries.
Install sensehat
dnf -y install zlib zlib-devel libjpeg-devel gcc gcc-c++ i2c-tools python3-devel python3 python3-pip cmake
pip3 install Cython Pillow numpy sense_hat smbus
Check the Sense Hat with i2cdetect
i2cdetect -y 1
[root@microshift ~]# i2cdetect -y 1
0 1 2 3 4 5 6 7 8 9 a b c d e f
00: -- -- -- -- -- -- -- --
10: -- -- -- -- -- -- -- -- -- -- -- -- 1c -- -- --
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
40: -- -- -- -- -- -- UU -- -- -- -- -- -- -- -- --
50: -- -- -- -- -- -- -- -- -- -- -- -- 5c -- -- 5f
60: -- -- -- -- -- -- -- -- -- -- 6a -- -- -- -- --
70: -- -- -- -- -- -- -- --
Install RTIMULib
dnf -y install git
cd ~
git clone https://github.com/RPi-Distro/RTIMULib.git
cd RTIMULib/
cd Linux/python
python3 setup.py build
python3 setup.py install
cd ../..
cd RTIMULib
mkdir build
cd build
cmake ..
make -j4
make install
ldconfig
# Optional test the sensors
cd /root/RTIMULib/Linux/RTIMULibDrive11
make -j4
make install
RTIMULibDrive11 # Ctrl-C to break
cd /root/RTIMULib/Linux/RTIMULibDrive10
make -j4
make install
RTIMULibDrive10 # Ctrl-C to break
# Optional
dnf -y install qt5-qtbase-devel
cd /root/RTIMULib/Linux/RTIMULibDemoGL
qmake-qt5
make -j4
make install
Test the SenseHat samples for the Sense Hat's LED matrix and sensors. With the latest version of sense_hat you will get “WARNING:root:Failed to initialise TCS34725 colour sensor. (sensor not present)” that you can ignore.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/sensehat-fedora-iot
# Enable random LEDs
python3 sparkles.py # Ctrl-C to interrupt
# Show multiple screens to test LEDs
python3 rainbow.py # Ctrl-C to interrupt
# First time you run the temperature.py, you may see “Temperature: 0 C”. Just run it again.
python3 temperature.py
# Show the Temperature, Pressure and Humidity
python3 testsensehat.py # Ctrl-C to interrupt
# Show two digits for multiple numbers
sed -i "s/32,32,32/255,255,255/" digits.py
python3 digits.py
# When a magnet gets close to SenseHAT, the LEDs will all turn red for 1/5 of a second
python3 magnetometer.py
# Find Magnetic North
python3 compass.py
Install MicroShift on the Raspberry Pi 4 AlmaLinux host
Setup crio and MicroShift Nightly CentOS Stream 9 aarch64
rpm -qi selinux-policy # selinux-policy-34.1.43
dnf -y install 'dnf-command(copr)'
curl https://copr.fedorainfracloud.org/coprs/g/redhat-et/microshift-nightly/repo/centos-stream-9/group_redhat-et-microshift-nightly-centos-stream-9.repo -o /etc/yum.repos.d/microshift-nightly-centos-stream-9.repo
cat /etc/yum.repos.d/microshift-nightly-centos-stream-9.repo
VERSION=1.24 # You can also use 1.21 or 1.25 with CentOS_9_Stream
curl -L -o /etc/yum.repos.d/devel:kubic:libcontainers:stable.repo https://download.opensuse.org/repositories/devel:/kubic:/libcontainers:/stable/Fedora_36/devel:kubic:libcontainers:stable.repo
curl -L -o /etc/yum.repos.d/devel:kubic:libcontainers:stable:cri-o:${VERSION}.repo https://download.opensuse.org/repositories/devel:kubic:libcontainers:stable:cri-o:${VERSION}/CentOS_8_Stream/devel:kubic:libcontainers:stable:cri-o:${VERSION}.repo
cat /etc/yum.repos.d/devel\:kubic\:libcontainers\:stable\:cri-o\:${VERSION}.repo
dnf -y install firewalld cri-o cri-tools microshift containernetworking-plugins # Be patient, this takes a few minutes
Install KVM on the host and validate the Host Virtualization Setup. The virt-host-validate command validates that the host is configured in a suitable way to run libvirt hypervisor driver qemu.
dnf -y install libvirt-client libvirt-nss qemu-kvm virt-manager virt-install virt-viewer
systemctl enable --now libvirtd
virt-host-validate qemu
Output:
[root@microshift ~]# systemctl enable --now libvirtd
Created symlink /etc/systemd/system/multi-user.target.wants/libvirtd.service → /usr/lib/systemd/system/libvirtd.service.
Created symlink /etc/systemd/system/sockets.target.wants/libvirtd.socket → /usr/lib/systemd/system/libvirtd.socket.
Created symlink /etc/systemd/system/sockets.target.wants/libvirtd-ro.socket → /usr/lib/systemd/system/libvirtd-ro.socket.
[root@microshift ~]# virt-host-validate qemu
QEMU: Checking if device /dev/kvm exists : PASS
QEMU: Checking if device /dev/kvm is accessible : PASS
QEMU: Checking if device /dev/vhost-net exists : PASS
QEMU: Checking if device /dev/net/tun exists : PASS
QEMU: Checking for cgroup 'cpu' controller support : PASS
QEMU: Checking for cgroup 'cpuacct' controller support : PASS
QEMU: Checking for cgroup 'cpuset' controller support : PASS
QEMU: Checking for cgroup 'memory' controller support : PASS
QEMU: Checking for cgroup 'devices' controller support : PASS
QEMU: Checking for cgroup 'blkio' controller support : PASS
QEMU: Checking for device assignment IOMMU support : WARN (Unknown if this platform has IOMMU support)
QEMU: Checking for secure guest support : WARN (Unknown if this platform has Secure Guest support)
Check that cni plugins are present
ls /opt/cni/bin/ # empty
ls /usr/libexec/cni # cni plugins
We will have systemd start and manage MicroShift. Refer to the microshift service for the three approaches.
systemctl enable --now crio microshift
# Copy flannel
# cp /opt/cni/bin/flannel /usr/libexec/cni/.
You may read about selecting zones for your interfaces.
systemctl enable firewalld --now
firewall-cmd --zone=trusted --add-source=10.42.0.0/16 --permanent
firewall-cmd --zone=public --add-port=80/tcp --permanent
firewall-cmd --zone=public --add-port=443/tcp --permanent
firewall-cmd --zone=public --add-port=5353/udp --permanent
firewall-cmd --reload
Additional ports may need to be opened. For external access to run kubectl or oc commands against MicroShift, add the 6443 port:
firewall-cmd --zone=public --permanent --add-port=6443/tcp
For access to services through NodePort, add the port range 30000-32767:
firewall-cmd --zone=public --permanent --add-port=30000-32767/tcp
firewall-cmd --reload
firewall-cmd --list-all --zone=public
firewall-cmd --get-default-zone
#firewall-cmd --set-default-zone=public
#firewall-cmd --get-active-zones
firewall-cmd --list-all
Check the microshift and crio logs
journalctl -u microshift -f
journalctl -u crio -f
The microshift service references the microshift binary in the /usr/bin directory
[root@microshift ~]# cat /usr/lib/systemd/system/microshift.service
[Unit]
Description=MicroShift
Wants=network-online.target crio.service
After=network-online.target crio.service
[Service]
WorkingDirectory=/usr/bin/
ExecStart=microshift run
Restart=always
User=root
[Install]
WantedBy=multi-user.target
Install the kubectl and the openshift oc client
ARCH=arm64
cd /tmp
dnf -y install tar
export OCP_VERSION=4.9.11 && \
curl -o oc.tar.gz https://mirror2.openshift.com/pub/openshift-v4/$ARCH/clients/ocp/$OCP_VERSION/openshift-client-linux-$OCP_VERSION.tar.gz && \
tar -xzvf oc.tar.gz && \
rm -f oc.tar.gz && \
install -t /usr/local/bin {kubectl,oc} && \
rm -f {README.md,kubectl,oc}
It will take around 3 minutes for all pods to start. Check the status of node and pods using kubectl or oc client.
export KUBECONFIG=/var/lib/microshift/resources/kubeadmin/kubeconfig
#watch "kubectl get nodes;kubectl get pods -A;crictl pods;crictl images"
watch "oc get nodes;oc get pods -A;crictl pods;crictl images"
You may also want to setup the Kubectl and oc autocomplete. The shell command-line completion allows you to quickly build your command without having to type every character.
dnf -y install bash-completion
source <(kubectl completion bash)
echo "source <(kubectl completion bash)" >> ~/.bashrc
source <(oc completion bash)
echo "source <(oc completion bash)" >> ~/.bashrc
and use use a shorthand alias for kubectl that also works with completion
alias k=kubectl
complete -o default -F __start_kubectl k
Install podman - We will use podman for containerized deployment of MicroShift and building images for the samples.
dnf -y install podman
Samples to run on MicroShift
We will run samples that will show the use of dynamic persistent volume, SenseHat and the USB camera.
1. InfluxDB/Telegraf/Grafana
The source code is available for this influxdb sample in github.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/influxdb
If you want to run all the steps in a single command, get the nodename.
oc get nodes
Output:
[root@microshift influxdb]# oc get nodes
NAME STATUS ROLES AGE VERSION
microshift.example.com Ready <none> 55m v1.21.0
Replace the annotation kubevirt.io/provisionOnNode with the above nodename and execute the runall-balena-dynamic.sh. Note that the node name is different when running MicroShift with the all-in-one containerized approach. So, you will use the microshift.example.com instead of the rpi.example.com.
sed -i "s|coreos|microshift.example.com|" influxdb-data-dynamic.yaml
sed -i "s|coreos|microshift.example.com|" grafana/grafana-data-dynamic.yaml
./runall-balena-dynamic.sh
We create and push the “measure:latest” image using the Dockerfile. The script will create a new project influxdb for this sample, install InfluxDB, install the pod for SenseHat measurements, install Telegraf and check the measurements for the telegraf database in InfluxDB. Finally, it will install Grafana.
This script will allocate dynamic persistent volumes using influxdb-data-dynamic.yaml and grafana-data-dynamic.yaml. The annotation provisionOnNode and the storageClassName are required for dynamic PV.
annotations:
kubevirt.io/provisionOnNode: microshift.example.com
spec:
storageClassName: kubevirt-hostpath-provisioner
Persistent Volumes and Claims Output:
[root@microshift influxdb]# oc get pv,pvc
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE
persistentvolume/pvc-8fc99c1b-1c7d-4af1-bcd6-745e102ab91f 57Gi RWO Delete Bound influxdb/influxdb-data kubevirt-hostpath-provisioner 3m42s
persistentvolume/pvc-e8fe1dee-ea97-448c-83fc-1ea006180d31 57Gi RWO Delete Bound influxdb/grafana-data kubevirt-hostpath-provisioner 63s
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/grafana-data Bound pvc-e8fe1dee-ea97-448c-83fc-1ea006180d31 57Gi RWO kubevirt-hostpath-provisioner 63s
persistentvolumeclaim/influxdb-data Bound pvc-8fc99c1b-1c7d-4af1-bcd6-745e102ab91f 57Gi RWO kubevirt-hostpath-provisioner 3m42s
Add the "<RaspberryPiIPAddress> grafana-service-influxdb.cluster.local" to /etc/hosts on your laptop and login to http://grafana-service-influxdb.cluster.local/login using admin/admin. You can change the password on first login or click on skip. Go to the Dashboards list (left menu > Dashboards > Manage). Open the Analysis Server dashboard to display monitoring information for MicroShift. Open the Balena Sense dashboard to show the temperature, pressure, and humidity from SenseHat.
Finally, after you are done working with this sample, you can run the deleteall-balena-dynamic.sh
./deleteall-balena-dynamic.sh
Deleting the persistent volume claims automatically deletes the persistent volumes.
2. Node Red live data dashboard with SenseHat sensor charts
We will install Node Red on the ARM device as a deployment within MicroShift, add the dashboard and view the gauges for temperature/pressure/humidity data from SenseHat on the dashboard.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/nodered
Build and push the arm64v8 image. I create and use the “karve/nodered:arm64”.
cd docker-custom/
# Replace docker with podman in docker-debian.sh, replace the tag
./docker-debian.sh
podman push karve/nodered:arm64
cd ..
Deploy Node Red with persistent volume for /data within the node red container
mkdir /var/hpvolumes/nodered
restorecon -R -v "/var/hpvolumes/*"
rm -rf /var/hpvolumes/nodered/*;cp -r nodered-volume/* /var/hpvolumes/nodered/.
oc new-project nodered
oc apply -f noderedpv.yaml -f noderedpvc.yaml -f nodered3.yaml -f noderedroute.yaml
oc get routes
oc -n nodered wait deployment nodered-deployment --for condition=Available --timeout=300s
oc logs deployment/nodered-deployment -f
Add the ipaddress of the Raspberry Pi 4 device for nodered-svc-nodered.cluster.local to /etc/hosts on your Laptop and browse to http://nodered-svc-nodered.cluster.local/
The following modules required for the dashboard have been preinstalled node-red-dashboard, node-red-node-smooth, node-red-node-pi-sense-hat. These can be seen under “Manage Palette - Install”. The Flow 1 or Flow 2 have already been imported from the nodered sample. This import to the Node Red can be done manually under “Import Nodes” and then click “Deploy”.
Double click the Sense HAT input node and make sure that all the events are checked. Select the Dashboard. Click on the down arrow on the top right and then the outward arrow below it in the tabs to view the sensor charts. You will see the Home by Default. You can see the state of the Joystick Up, Down, Left, Right or Pressed. Click on the Hamburger Menu (3 lines) and select PiSenseHAT. If you selected the Flow 1, you could click on the Input for the Timestamp under “Dot Matrix” to see the “Alarm” message scroll on the SenseHat LED. You can see the screenshots for these dashboards in previous blogs.
We can continue running the next sample that will reuse this Node Red deployment. If the Node Red Deployment is no longer required, we can delete it as follows:
cd ~/microshift/raspberry-pi/nodered
oc delete -f noderedpv.yaml -f noderedpvc.yaml -f nodered3.yaml -f noderedroute.yaml -n nodered
oc project default
oc delete project nodered
3. TensorFlow Lite Python object detection example in MicroShift with SenseHat and Node Red
This example requires the same Node Red setup as in the previous Sample 2.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/object-detection
We will build the image for object detection send pictures and web socket chat messages to Node Red when a person is detected using a pod in microshift.
podman build -t docker.io/karve/object-detection-raspberrypi4 .
podman push docker.io/karve/object-detection-raspberrypi4:latest
Update the env WebSocketURL and ImageUploadURL in object-detection.yaml as shown below. Also update the hostAliases in object-detection.yaml to point to your raspberry pi 4 ip address (192.168.1.209 shown below).
env:
- name: WebSocketURL
value: "ws://nodered-svc-nodered.cluster.local/ws/chat"
- name: ImageUploadURL
value: http://nodered-svc-nodered.cluster.local/upload
hostAliases:
- hostnames:
- nodered-svc-nodered.cluster.local
ip: 192.168.1.209
Create the deployment
oc project default
oc apply -f object-detection.yaml
oc -n default wait deployment object-detection-deployment --for condition=Available --timeout=300s
We will see pictures being sent to Node Red when a person is detected at http://nodered-svc-nodered.cluster.local/#flow/3e30dc50ae28f61f and chat messages at http://nodered-svc-nodered.cluster.local/chat. When we are done testing, we can delete the deployment
oc delete -f object-detection.yaml
4. Running a Virtual Machine Instance on MicroShift
Find the latest version of the KubeVirt Operator and install.
LATEST=$(curl -L https://storage.googleapis.com/kubevirt-prow/devel/nightly/release/kubevirt/kubevirt/latest-arm64)
echo $LATEST
#LATEST=20221123 # I used this version
oc apply -f https://storage.googleapis.com/kubevirt-prow/devel/nightly/release/kubevirt/kubevirt/${LATEST}/kubevirt-operator-arm64.yaml
oc apply -f https://storage.googleapis.com/kubevirt-prow/devel/nightly/release/kubevirt/kubevirt/${LATEST}/kubevirt-cr-arm64.yaml
oc adm policy add-scc-to-user privileged -n kubevirt -z kubevirt-operator
# The .status.phase will show Deploying multiple times and finally Deployed
oc get kubevirt.kubevirt.io/kubevirt -n kubevirt -o=jsonpath="{.status.phase}" -w # Ctrl-C to break
oc -n kubevirt wait kv kubevirt --for condition=Available --timeout=300s
oc get pods -n kubevirt
We can build the OKD Web Console (Codename: “bridge”) from the source as mentioned in Part 9. We will run the “bridge” as a container image that we run within MicroShift.
cd /root/microshift/raspberry-pi/console
oc create serviceaccount console -n kube-system
oc create clusterrolebinding console --clusterrole=cluster-admin --serviceaccount=kube-system:console -n kube-system
sleep 5
oc get serviceaccount console --namespace=kube-system -o jsonpath='{.secrets[0].name}' | grep console-token
oc get serviceaccount console --namespace=kube-system -o jsonpath='{.secrets[1].name}' | grep console-token
Replace BRIDGE_K8S_MODE_OFF_CLUSTER_ENDPOINT value https://192.168.1.209:6443 with your raspberry pi 4's ip address, and secretRef token with the console-token-* from above two secret names for BRIDGE_K8S_AUTH_BEARER_TOKEN in okd-web-console-install.yaml. Then apply/create the okd-web-console-install.yaml.
vi okd-web-console-install.yaml
oc apply -f okd-web-console-install.yaml
oc expose svc console-np-service -n kube-system
oc get routes -n kube-system
oc -n kube-system wait deployment console-deployment --for condition=Available --timeout=300s
oc logs deployment/console-deployment -f -n kube-system
Add the Raspberry Pi IP address to /etc/hosts on your Macbook Pro to resolve console-np-service-kube-system.cluster.local. Now you can access the OKD Web Console from your Laptop http://console-np-service-kube-system.cluster.local/. If you see a blank page, you probably have the value of BRIDGE_K8S_MODE_OFF_CLUSTER_ENDPOINT set incorrectly.
We can optionally preload the fedora image into crio (if using the all-in-one containerized approach, this needs to be run within the microshift pod running in podman)
crictl pull quay.io/kubevirt/fedora-cloud-container-disk-demo:20210811_9fec1f849-arm64
Now let’s create a Fedora Virtual Machine Instance using the vmi-fedora.yaml.
cd /root/microshift/raspberry-pi/vmi
oc apply -f vmi-fedora.yaml
watch oc get vmi,pods
The output for the virtualmachineinstance PHASE goes from “Scheduling” to “Scheduled” to “Running” after the virt-launcher-vmi-fedora pod STATUS goes from “Init” to “Running”. Note down the ip address of the vmi-fedora Virtual Machine Instance. Directly connect to the VMI from the Raspberry Pi 4 with fedora as the user and password. Note that it will take another minute or two after the VMI goes to Running state to ssh to the instance. You can look at the console (GUI) or use virtctl (third way below) to see if the VMI is ready.
Output:
[root@microshift vmi]# oc get vmi
NAME AGE PHASE IP NODENAME READY
vmi-fedora 2m38s Running 10.42.0.19 microshift.example.com True
[root@microshift vmi]# ssh fedora@10.42.0.19 ping -c 2 google.com
ssh: connect to host 10.42.0.19 port 22: No route to host
[root@microshift vmi]# ssh fedora@10.42.0.19 ping -c 2 google.com
The authenticity of host '10.42.0.19 (10.42.0.19)' can't be established.
ED25519 key fingerprint is SHA256:YGTHSuTB+I2v+vxd/7OfF5INarCZNzXmrzHvMlhrFdI.
This key is not known by any other names
Are you sure you want to continue connecting (yes/no/[fingerprint])? yes
Warning: Permanently added '10.42.0.19' (ED25519) to the list of known hosts.
fedora@10.42.0.19's password:
PING google.com (142.250.65.174) 56(84) bytes of data.
64 bytes from lga25s71-in-f14.1e100.net (142.250.65.174): icmp_seq=1 ttl=59 time=4.70 ms
64 bytes from lga25s71-in-f14.1e100.net (142.250.65.174): icmp_seq=2 ttl=59 time=4.69 ms
--- google.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 4.690/4.693/4.696/0.003 ms
Alternatively, a second way is to create a Pod to run the ssh client and connect to the Fedora VM from this pod. Let’s create that openssh-client pod:
oc run alpine --privileged --rm -ti --image=alpine -- /bin/sh
apk update && apk add --no-cache openssh-client
or
oc run sshclient --privileged --rm -ti --image=karve/alpine-sshclient:arm64 -- /bin/sh
#oc attach sshclient -c sshclient -i -t
Then, ssh to the Fedora VMI from this openssh-client container.
Output:
[root@microshift vmi]# oc run alpine --privileged --rm -ti --image=alpine -- /bin/sh
If you don't see a command prompt, try pressing enter.
/ # apk update && apk add --no-cache openssh-client
fetch https://dl-cdn.alpinelinux.org/alpine/v3.17/main/aarch64/APKINDEX.tar.gz
fetch https://dl-cdn.alpinelinux.org/alpine/v3.17/community/aarch64/APKINDEX.tar.gz
v3.17.0_rc4-69-gfea827d99a [https://dl-cdn.alpinelinux.org/alpine/v3.17/main]
v3.17.0-6-gba64c76844 [https://dl-cdn.alpinelinux.org/alpine/v3.17/community]
OK: 17672 distinct packages available
fetch https://dl-cdn.alpinelinux.org/alpine/v3.17/main/aarch64/APKINDEX.tar.gz
fetch https://dl-cdn.alpinelinux.org/alpine/v3.17/community/aarch64/APKINDEX.tar.gz
(1/6) Installing openssh-keygen (9.1_p1-r1)
(2/6) Installing ncurses-terminfo-base (6.3_p20221119-r0)
(3/6) Installing ncurses-libs (6.3_p20221119-r0)
(4/6) Installing libedit (20221030.3.1-r0)
(5/6) Installing openssh-client-common (9.1_p1-r1)
(6/6) Installing openssh-client-default (9.1_p1-r1)
Executing busybox-1.35.0-r29.trigger
OK: 13 MiB in 21 packages
/ # ssh fedora@10.42.0.19 "bash -c \"ping -c 2 google.com\""
The authenticity of host '10.42.0.19 (10.42.0.19)' can't be established.
ED25519 key fingerprint is SHA256:YGTHSuTB+I2v+vxd/7OfF5INarCZNzXmrzHvMlhrFdI.
This key is not known by any other names.
Are you sure you want to continue connecting (yes/no/[fingerprint])? yes
Warning: Permanently added '10.42.0.19' (ED25519) to the list of known hosts.
fedora@10.42.0.19's password:
PING google.com (142.250.65.238) 56(84) bytes of data.
64 bytes from lga25s73-in-f14.1e100.net (142.250.65.238): icmp_seq=1 ttl=117 time=3.89 ms
64 bytes from lga25s73-in-f14.1e100.net (142.250.65.238): icmp_seq=2 ttl=117 time=3.88 ms
--- google.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 3.882/3.885/3.889/0.003 ms
/ # exit
Session ended, resume using 'oc attach alpine -c alpine -i -t' command when the pod is running
pod "alpine" deleted
A third way to connect to the VM is to use the virtctl console. You can compile your own virtctl as was described in Part 9. To simplify, we copy virtctl arm64 binary from prebuilt container image to /usr/local/bin on the Raspberry Pi 4 and connect to the VMI using “virtctl console” command.
id=$(podman create docker.io/karve/kubevirt:arm64)
podman cp $id:_out/cmd/virtctl/virtctl /usr/local/bin
podman rm -v $id
virtctl console vmi-fedora
Output:
[root@microshift vmi]# id=$(podman create docker.io/karve/kubevirt:arm64)
Trying to pull docker.io/karve/kubevirt:arm64...
Getting image source signatures
Copying blob 7065f6098427 done
Copying config 1c7a5aa443 done
Writing manifest to image destination
Storing signatures
[root@microshift vmi]# podman cp $id:_out/cmd/virtctl/virtctl /usr/local/bin
[root@microshift vmi]# podman rm -v $id
fcd356f9c9e99fbd6dc49d39707a94c61d2ca4e9ca75ac7692fcaa7dabac0987
[root@microshift vmi]# virtctl console vmi-fedora
Successfully connected to vmi-fedora console. The escape sequence is ^]
vmi-fedora login: fedora
Password:
[fedora@vmi-fedora ~]$ ping -c 2 ibm.com
PING ibm.com (184.87.72.96) 56(84) bytes of data.
64 bytes from a184-87-72-96.deploy.static.akamaitechnologies.com (184.87.72.96): icmp_seq=1 ttl=57 time=6.04 ms
64 bytes from a184-87-72-96.deploy.static.akamaitechnologies.com (184.87.72.96): icmp_seq=2 ttl=57 time=4.63 ms
--- ibm.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1001ms
rtt min/avg/max/mdev = 4.627/5.335/6.044/0.708 ms
[fedora@vmi-fedora ~]$ # ^] to detach
[root@microshift vmi]#
When done, we can delete the VMI
oc delete -f vmi-fedora.yaml
We can run other VM and VMI samples for alpine, cirros and fedora images as in Part 9. You may continue to the next sample 4 where we create a new CentOS VM image. When done, you may delete kubevirt operator
oc delete -f https://storage.googleapis.com/kubevirt-prow/devel/nightly/release/kubevirt/kubevirt/${LATEST}/kubevirt-cr-arm64.yaml
oc delete -f https://storage.googleapis.com/kubevirt-prow/devel/nightly/release/kubevirt/kubevirt/${LATEST}/kubevirt-operator-arm64.yaml
5. Running a Virtual Machine on MicroShift
We will create a Virtual Machine (VM) using the vm-centos9.yaml. When a VMI is owned by a VM or by another object, it is managed through its owner in the web console or by using the oc command-line interface (CLI). A VirtualMachine provides additional management capabilities to a VirtualMachineInstance inside the cluster. A VirtualMachine will make sure that a VirtualMachineInstance object with an identical name will be present in the cluster, if spec.running is set to true. Further it will make sure that a VirtualMachineInstance will be removed from the cluster if spec.running is set to false. Once a VirtualMachineInstance is created, its state will be tracked via status.created and status.ready fields of the VirtualMachine.
We need to first create an image for use in vm-centos9.yaml. I have already created the docker.io/karve/centos-stream-genericcloud-9-20221206:arm64 as follows:
a. Download the required/latest image from https://cloud.centos.org/altarch/9-stream/aarch64/images/. I used the https://cloud.centos.org/altarch/9-stream/aarch64/images/CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 image.
b. Create the following Dockerfile. You can do this on your Laptop/Macbook.
FROM docker.io/kubevirt/container-disk-v1alpha
ADD CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 /disk/
Build and push the image to repository
podman build -t docker.io/karve/centos-stream-genericcloud-9-20221206:arm64 .
podman push docker.io/karve/centos-stream-genericcloud-9-20221206:arm64
Now let’s create and start the CentOS 9 Stream VM on the Raspberry Pi 4. Optionally, pull the image using crictl.
crictl pull docker.io/karve/centos-stream-genericcloud-9-20221206:arm64
cd /root/microshift/raspberry-pi/vmi
Update the vm-centos9.yaml with your public key in ssh_authorized_keys so you can ssh
vi vm-centos9.yaml
oc apply -f vm-centos9.yaml
virtctl start vm-centos9
oc get vm,vmi
virtctl console vm-centos9
Output:
[root@microshift vmi]# cd /root/microshift/raspberry-pi/vmi
[root@microshift vmi]# oc apply -f vm-centos9.yaml
virtualmachine.kubevirt.io/vm-centos9 created
[root@microshift vmi]# oc get vm
NAME AGE STATUS READY
vm-centos9 18s Stopped False
[root@microshift vmi]# virtctl start vm-centos9
VM vm-centos9 was scheduled to start
[root@microshift vmi]# oc get vm,vmi
NAME AGE STATUS READY
virtualmachine.kubevirt.io/vm-centos9 2m34s Starting False
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/vm-centos9 1s Pending
[root@microshift vmi]# oc get pods
NAME READY STATUS RESTARTS AGE
virt-launcher-vm-centos9-lv4vb 2/2 Running 0 56s
[root@microshift vmi]# oc get vm,vmi
NAME AGE STATUS READY
virtualmachine.kubevirt.io/vm-centos9 3m47s Running True
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/vm-centos9 74s Running 10.42.0.14 microshift.example.com True
[root@microshift vmi]# virtctl console vm-centos9
Successfully connected to vm-centos9 console. The escape sequence is ^]
… [You will see the output as the system boots up]
CentOS Stream 9
Kernel 5.14.0-206.el9.aarch64 on an aarch64
Activate the web console with: systemctl enable --now cockpit.socket
vm-centos9 login: cloud-user
Password:
[cloud-user@vm-centos9 ~]$ ping -c 2 google.com
PING google.com (142.251.32.110) 56(84) bytes of data.
64 bytes from lga25s77-in-f14.1e100.net (142.251.32.110): icmp_seq=1 ttl=117 time=9.91 ms
64 bytes from lga25s77-in-f14.1e100.net (142.251.32.110): icmp_seq=2 ttl=117 time=5.24 ms
--- google.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1001ms
rtt min/avg/max/mdev = 5.244/7.576/9.909/2.332 ms
[cloud-user@vm-centos9 ~]$ # ^] to exit
[root@microshift vmi]#
We can also ssh to the IP address of the VM as the root (using the ssh private key) or as cloud-user using the password centos. Using the Web Console, we can see the VMs and VMIs in a single list view. VMIs are not surfaced in the list if they are owned by a VM. We can also execute the Actions from the UI.
When done, we can stop the Virtual Machine Instance and delete the Virtual Machine.
virtctl stop vm-centos9
oc delete -f vm-centos9.yaml
With MicroShift Virtualization, we let VMs run as a native citizen next to container workloads. In addition to exporting the ssh as a service, we can expose other services in the project the VM is running. Deployed in the same project both use the same network and can access each other’s services.
6. Containerized Data Importer (CDI)
CDI is a utility designed to import Virtual Machine images for use with Kubevirt. At a high level, a PersistentVolumeClaim (PVC) is created. A custom controller watches for importer specific claims, and when discovered, starts an import process to create a raw image with the desired content into the associated PVC.
Update the volume.kubernetes.io/selected-node to your node name (microshift.example.com) in centos9-pv.yaml.
Check the latest arm64 version at https://quay.io/repository/kubevirt/cdi-operator?tab=tags&tag=latest
VERSION=v1.55.2
ARM_VERSION=20221210_a6ebd75e-arm64 # Use the arm64 tag from https://quay.io/repository/kubevirt/cdi-operator?tab=tags&tag=latest
# The version does not work with arm64 images
# oc apply -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-operator.yaml
# So we use the ARM_VERSION
curl -sL https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-operator.yaml | sed "s/$VERSION/$ARM_VERSION/g" | oc apply -f -
# Wait for cdi-operator to start
# Next create the cdi-cr that will create the apiserver, deployment and uploadproxy
oc apply -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-cr.yaml
oc get apiservices
oc api-resources --api-group=cdi.kubevirt.io
cd ~/microshift/raspberry-pi/vmi
oc apply -f centos9-dv.yaml # Create a persistent volume by downloading the CentOS image
oc apply -f vm-centos9-datavolume.yaml # Create a CentOS VM by cloning the persistent volume with the above CentOS image
Output:
[root@microshift vmi]# curl -sL https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-operator.yaml | sed "s/$VERSION/$ARM_VERSION/g" | oc apply -f -
namespace/cdi created
customresourcedefinition.apiextensions.k8s.io/cdis.cdi.kubevirt.io created
clusterrole.rbac.authorization.k8s.io/cdi-operator-cluster created
clusterrolebinding.rbac.authorization.k8s.io/cdi-operator created
serviceaccount/cdi-operator created
role.rbac.authorization.k8s.io/cdi-operator created
rolebinding.rbac.authorization.k8s.io/cdi-operator created
deployment.apps/cdi-operator created
configmap/cdi-operator-leader-election-helper created
[root@microshift vmi]# oc get deployments,pods -n cdi
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/cdi-operator 1/1 1 1 19s
NAME READY STATUS RESTARTS AGE
pod/cdi-operator-794d4f98fc-hswtk 1/1 Running 0 18s
[root@microshift vmi]# oc apply -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-cr.yaml
cdi.cdi.kubevirt.io/cdi created
[root@microshift vmi]# oc get deployments,pods -n cdi
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/cdi-apiserver 1/1 1 1 88s
deployment.apps/cdi-deployment 1/1 1 1 85s
deployment.apps/cdi-operator 1/1 1 1 2m46s
deployment.apps/cdi-uploadproxy 1/1 1 1 82s
NAME READY STATUS RESTARTS AGE
pod/cdi-apiserver-6679949df6-7lzc8 1/1 Running 0 87s
pod/cdi-deployment-798bd75486-6chlm 1/1 Running 0 85s
pod/cdi-operator-794d4f98fc-hswtk 1/1 Running 0 2m45s
pod/cdi-uploadproxy-7fb8544d88-574xj 1/1 Running 0 81s
Notice the cdi-apiserver logs that show the "no valid subject specified". We need to change the requestheader-allowed-names ~/microshift/pkg/controllers/kube-apiserver.go. We will fix this issue later by setting the requestheader-allowed-names="". This blank option indicates to an extension apiserver that any CN is acceptable.
[root@microshift vmi]# oc logs deployment/cdi-apiserver -f
2022/12/10 11:31:16 http: TLS handshake error from 10.42.0.1:51380: no valid subject specified
2022/12/10 11:31:16 http: TLS handshake error from 10.42.0.1:51404: no valid subject specified
…
The microshift logs show that it is trying to get a response directly from the pod ip address
Dec 10 13:41:49 microshift.example.com microshift[601]: E1210 13:41:49.572183 601 available_controller.go:508] v1beta1.upload.cdi.kubevirt.io failed with: failing or missing response from https://10.42.0.7:8443/apis/upload.cdi.kubevirt.io/v1beta1: Get "https://10.42.0.7:8443/apis/upload.cdi.kubevirt.io/v1beta1": remote error: tls: bad certificate
Dec 10 13:41:49 microshift.example.com microshift[601]: E1210 13:41:49.958797 601 available_controller.go:508] v1alpha1.upload.cdi.kubevirt.io failed with: failing or missing response from https://10.42.0.7:8443/apis/upload.cdi.kubevirt.io/v1alpha1: Get "https://10.42.0.7:8443/apis/upload.cdi.kubevirt.io/v1alpha1": remote error: tls: bad certificate
[root@microshift ~]# kubectl api-resources --api-group=cdi.kubevirt.io
NAME SHORTNAMES APIVERSION NAMESPACED KIND
cdiconfigs cdi.kubevirt.io/v1beta1 false CDIConfig
cdis cdi,cdis cdi.kubevirt.io/v1beta1 false CDI
dataimportcrons dic,dics cdi.kubevirt.io/v1beta1 true DataImportCron
datasources das cdi.kubevirt.io/v1beta1 true DataSource
datavolumes dv,dvs cdi.kubevirt.io/v1beta1 true DataVolume
objecttransfers ot,ots cdi.kubevirt.io/v1beta1 false ObjectTransfer
storageprofiles cdi.kubevirt.io/v1beta1 false StorageProfile
error: unable to retrieve the complete list of server APIs: upload.cdi.kubevirt.io/v1alpha1: the server is currently unable to handle the request, upload.cdi.kubevirt.io/v1beta1: the server is currently unable to handle the request
Update the kubevirt.io/provisionOnNode in centos-dv.yaml with your microshift node name.
[root@microshift vmi]# oc apply -f centos9-dv.yaml
datavolume.cdi.kubevirt.io/centos9-dv created
# This will create the pod/importer-centos9-dv to import the image
[root@microshift vmi]# oc get dv,pvc,pods,vmi
NAME PHASE PROGRESS RESTARTS AGE
datavolume.cdi.kubevirt.io/centos9-dv ImportInProgress N/A 12s
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 12s
NAME READY STATUS RESTARTS AGE
pod/importer-centos9-dv 1/1 Running 0 12s
[root@microshift vmi]# oc get dv -w
NAME PHASE PROGRESS RESTARTS AGE
centos9-dv ImportInProgress 8.19% 69s
centos9-dv ImportInProgress 9.19% 70s
…
centos9-dv ImportInProgress 99.71% 14m
centos9-dv Succeeded 100.0% 14m
centos9-dv Succeeded 100.0% 14m
[root@microshift vmi]# oc get dv,pvc,pods,vmi
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 15m
Update the persistent volume claim annotation kubevirt.io/provisionOnNode to your microshift node name, update with your public key and apply the vm-centos9-datavolume.yaml.
[root@microshift vmi]# oc apply -f vm-centos9-datavolume.yaml
virtualmachine.kubevirt.io/centos9instance1 created
[root@microshift vmi]# watch oc get dv,pvc,pods,vmi
NAME PHASE PROGRESS RESTARTS AGE
datavolume.cdi.kubevirt.io/centos9instance1 CloneInProgress N/A 37s
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS
AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner
16m
persistentvolumeclaim/centos9instance1 Bound pvc-efd3b974-32f9-4357-bbb0-df6103445435 57Gi RWO kubevirt-hostpath-provisioner
37s
NAME READY STATUS RESTARTS AGE
pod/cdi-upload-centos9instance1 1/1 Running 0 32s
pod/efd3b974-32f9-4357-bbb0-df6103445435-source-pod 1/1 Running 0 21s
[root@microshift vmi]# oc get dv -w
NAME PHASE PROGRESS RESTARTS AGE
centos9instance1 CloneInProgress 0.98% 70s
centos9instance1 CloneInProgress 1.02% 70s
centos9instance1 CloneInProgress 1.12% 70s
…
centos9instance1 CloneInProgress 100.00% 5m26s
centos9instance1 Succeeded 100.0% 5m28s
centos9instance1 Succeeded 100.0% 5m29s
[root@microshift vmi]# oc get dv,pvc,pods,vmi
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 22m
persistentvolumeclaim/centos9instance1 Bound pvc-efd3b974-32f9-4357-bbb0-df6103445435 57Gi RWO kubevirt-hostpath-provisioner 6m27s
NAME READY STATUS RESTARTS AGE
pod/virt-launcher-centos9instance1-5b4xk 1/1 Running 0 59s
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/centos9instance1 59s Running 10.42.0.34 microshift.example.com True
Connect to the VM console using virtctl
[root@microshift vmi]# virtctl console centos9instance1
Successfully connected to centos9instance1 console. The escape sequence is ^]
EFI stub: Booting Linux Kernel...
EFI stub: EFI_RNG_PROTOCOL unavailable
EFI stub: Using DTB from configuration table
EFI stub: Exiting boot services...
[ 0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd083]
[ 0.000000] Linux version 5.14.0-206.el9.aarch64 (mockbuild@aarch64-01.stream.rdu2.redhat.com) (gcc (GCC) 11.3.1 20221121 (Red Hat 11.3.1-4), GNU ld version 2.35.2-24.el9) #1 SMP PREEMPT_DYNAMIC Mon Dec 5 09:08:53 UTC 2022
[ 0.000000] The list of certified hardware and cloud instances for Red Hat Enterprise Linux 9 can be viewed at the Red Hat Ecosystem Catalog, https://catalog.redhat.com.
[ 0.000000] efi: EFI v2.70 by EDK II
[ 0.000000] efi: SMBIOS 3.0=0x7f5b0000 MEMATTR=0x7e755018 ACPI 2.0=0x7c030018 MEMRESERVE=0x7c223e18
…
CentOS Stream 9
Kernel 5.14.0-206.el9.aarch64 on an aarch64
Activate the web console with: systemctl enable --now cockpit.socket
centos9instance1 login: cloud-user
Password:
[cloud-user@centos9instance1 ~]$ ping google.com
PING google.com (142.250.65.174) 56(84) bytes of data.
64 bytes from lga25s71-in-f14.1e100.net (142.250.65.174): icmp_seq=1 ttl=59 time=23.9 ms
64 bytes from lga25s71-in-f14.1e100.net (142.250.65.174): icmp_seq=2 ttl=59 time=4.80 ms
--- google.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 4.795/14.366/23.937/9.571 ms
[cloud-user@centos9instance1 ~]$ # ^] to exit
[root@microshift vmi]#
Directly ssh to the VMI as cloud-user using your private key
[root@microshift vmi]# ssh -i ~/amazontestkey.pem cloud-user@10.42.0.34
The authenticity of host '10.42.0.34 (10.42.0.34)' can't be established.
ED25519 key fingerprint is SHA256:C6Yr+wIm7gbinR7pr9pT2kWIostw8c9fwCa9QBWj7I0.
This key is not known by any other names
Are you sure you want to continue connecting (yes/no/[fingerprint])? yes
Warning: Permanently added '10.42.0.34' (ED25519) to the list of known hosts.
Last login: Fri Dec 9 18:48:06 2022
[cloud-user@centos9instance1 ~]$ cat /etc/os-release
NAME="CentOS Stream"
VERSION="9"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="9"
PLATFORM_ID="platform:el9"
PRETTY_NAME="CentOS Stream 9"
ANSI_COLOR="0;31"
LOGO="fedora-logo-icon"
CPE_NAME="cpe:/o:centos:centos:9"
HOME_URL="https://centos.org/"
BUG_REPORT_URL="https://bugzilla.redhat.com/"
REDHAT_SUPPORT_PRODUCT="Red Hat Enterprise Linux 9"
REDHAT_SUPPORT_PRODUCT_VERSION="CentOS Stream"
[cloud-user@centos9instance1 ~]$ uname -a
Linux centos9instance1 5.14.0-206.el9.aarch64 #1 SMP PREEMPT_DYNAMIC Mon Dec 5 09:08:53 UTC 2022 aarch64 aarch64 aarch64 GNU/Linux
[cloud-user@centos9instance1 ~]$ exit
logout
Connection to 10.42.0.34 closed.
Directly ssh to the VMI as root using your private key
[root@microshift vmi]# ssh -i ~/amazontestkey.pem root@10.42.0.34
Activate the web console with: systemctl enable --now cockpit.socket
[root@centos9instance1 ~]# df -h
Filesystem Size Used Avail Use% Mounted on
devtmpfs 4.0M 0 4.0M 0% /dev
tmpfs 351M 0 351M 0% /dev/shm
tmpfs 141M 4.1M 137M 3% /run
/dev/vda2 9.5G 1.2G 8.3G 13% /
/dev/vda1 599M 7.0M 592M 2% /boot/efi
tmpfs 71M 0 71M 0% /run/user/1000
tmpfs 71M 0 71M 0% /run/user/0
[root@centos9instance1 ~]# exit
logout
Connection to 10.42.0.34 closed.
[root@microshift vmi]# oc get vm,vmi,pvc
NAME AGE STATUS READY
virtualmachine.kubevirt.io/centos9instance1 19m Running True
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/centos9instance1 14m Running 10.42.0.34 microshift.example.com True
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 35m
persistentvolumeclaim/centos9instance1 Bound pvc-efd3b974-32f9-4357-bbb0-df6103445435 57Gi RWO kubevirt-hostpath-provisioner 19m
We can delete the Virtual Machine, the VMI and the persistent volume
[root@microshift vmi]# oc delete -f vm-centos9-datavolume.yaml
virtualmachine.kubevirt.io "centos9instance1" deleted
Or, we can delete separately without cascading
[root@microshift vmi]# oc delete vm centos9instance1 --cascade=false
warning: --cascade=false is deprecated (boolean value) and can be replaced with --cascade=orphan.
virtualmachine.kubevirt.io "centos9instance1" deleted
[root@microshift vmi]# oc get vm,vmi,pvc
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/centos9instance1 14m Running 10.42.0.34 microshift.example.com True
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 36m
persistentvolumeclaim/centos9instance1 Bound pvc-efd3b974-32f9-4357-bbb0-df6103445435 57Gi RWO kubevirt-hostpath-provisioner 20m
[root@microshift vmi]# oc delete vmi centos9instance1 --cascade=orphan
virtualmachineinstance.kubevirt.io "centos9instance1" deleted
[root@microshift vmi]# oc get vm,vmi,pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 36m
persistentvolumeclaim/centos9instance1 Bound pvc-efd3b974-32f9-4357-bbb0-df6103445435 57Gi RWO kubevirt-hostpath-provisioner 20m
[root@microshift vmi]# oc delete pvc centos9instance1
persistentvolumeclaim "centos9instance1" deleted
[root@microshift vmi]# oc get vm,vmi,pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
persistentvolumeclaim/centos9-dv Bound pvc-084d9711-410c-4ef8-8614-66b73f96ecb1 57Gi RWO kubevirt-hostpath-provisioner 36m
7. Virtctl image-upload with CDI
DataVolume objects are custom resources that are provided by the Containerized Data Importer (CDI) project. Data volumes orchestrate import, clone, and upload operations that are associated with an underlying persistent volume claim (PVC). Data volumes are integrated with OKD Virtualization, and they prevent a virtual machine from being started before the PVC has been prepared. The Containerized Data Importer (CDI) requires scratch space (temporary storage) to complete some operations, such as importing and uploading virtual machine images. During this process, CDI provisions a scratch space PVC equal to the size of the PVC backing the destination data volume (DV). The scratch space PVC is deleted after the operation completes or aborts. The CDI supported operations matrix shows the supported CDI operations for content types against endpoints, and which of these operations requires scratch space (qcow2 or raw with gz or xz compression are supported, do not use the tar.gz or tar.xz).
Add the ipaddress of your Raspberry Pi 4 for the cdi-uploadproxy-cdi.cluster.local to /etc/hosts (on the Raspberry Pi 4) and update the kubevirt.io/provisionOnNode in example-upload-dv.yaml with your microshift node name. Note that it creates a pod for uploading the image and two volumes, one of them is a scratch volume. To allow insecure server connections when using HTTPS, use the --insecure parameter. Be aware that when you use the --insecure flag, the authenticity of the upload endpoint is not verified.
echo $ipaddress cdi-uploadproxy-cdi.cluster.local >> /etc/hosts
oc apply -f example-upload-dv.yaml
virtctl image-upload dv example-upload-dv --namespace default --size 10Gi --image-path `pwd`/CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 --wait-secs 1200 --no-create --uploadproxy-url=https://cdi-uploadproxy-cdi.cluster.local --insecure
Output:
[root@microshift vmi]# oc apply -f example-upload-dv.yaml
datavolume.cdi.kubevirt.io/example-upload-dv created
[root@microshift vmi]# watch oc get dv,pvc,pods -n default
NAME PHASE PROGRESS RESTARTS AGE
datavolume.cdi.kubevirt.io/example-upload-dv UploadScheduled N/A 41s
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS
AGE
persistentvolumeclaim/example-upload-dv Bound pvc-a36a235b-3683-44d6-ba1e-9754155f632f 57Gi RWO kubevirt-hostpath-provi
sioner 40s
persistentvolumeclaim/example-upload-dv-scratch Bound pvc-4d836e63-f705-4ca0-9f89-c44566f3a2f6 57Gi RWO kubevirt-hostpath-provi
sioner 38s
NAME READY STATUS RESTARTS AGE
pod/cdi-upload-example-upload-dv 0/1 Running 0 38s
The default microshift binary will give the following error with image-upload
[root@microshift vmi]# virtctl image-upload dv example-upload-dv --namespace default --size 10Gi --image-path `pwd`/CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 --wait-secs 1200 --no-create --uploadproxy-url=https://cdi-uploadproxy-cdi.cluster.local --insecure
Using existing PVC default/example-upload-dv
Uploading data to https://cdi-uploadproxy-cdi.cluster.local
the server is currently unable to handle the request (post uploadtokenrequests.upload.cdi.kubevirt.io)
We need to fix this. Modify the ~/microshift/pkg/controllers/kube-apiserver.go and update the line requestheader-allowed-names to empty and recompile microshift binary.
"--requestheader-allowed-names=",
Then run the cleanup, replace the /usr/bin/microshift binary and restart microshift.
[root@microshift hack]# ./cleanup.sh
DATA LOSS WARNING: Do you wish to stop and cleanup ALL MicroShift data AND cri-o container workloads?
1) Yes
2) No
#? 1
Stopping microshift
Removing crio pods
…
[root@microshift hack]# cd ..
[root@microshift microshift]# vi pkg/controllers/kube-apiserver.go
[root@microshift microshift]# make
[root@microshift microshift]# cp microshift /usr/bin/microshift
cp: overwrite '/usr/bin/microshift'? y
[root@microshift microshift]# restorecon -R -v /usr/bin/microshift
[root@microshift microshift]# systemctl start microshift
If you get 503 Service Unavailable in the microshift journalctl logs, just run “systemctl restart microshift”. Then, install the CDI as shown previously and create the example-upload-dv.yaml.
[root@microshift microshift]# cd raspberry-pi/vmi
[root@microshift vmi]# oc apply -f example-upload-dv.yaml
datavolume.cdi.kubevirt.io/example-upload-dv created
[root@microshift vmi]# virtctl image-upload dv example-upload-dv --namespace default --size 10Gi --image-path `pwd`/CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 --wait-secs 1200 --no-create --uploadproxy-url=https://cdi-uploadproxy-cdi.cluster.local --insecure
Using existing PVC default/example-upload-dv
Uploading data to https://cdi-uploadproxy-cdi.cluster.local
748.12 MiB / 748.12 MiB [==============================================================================================================] 100.00% 1m38s
Uploading data completed successfully, waiting for processing to complete, you can hit ctrl-c without interrupting the progress
Processing completed successfully
Uploading /root/microshift/raspberry-pi/vmi/CentOS-Stream-GenericCloud-9-20221206.0.aarch64.qcow2 completed successfully
The upload pod logs show the image being processed as it is uploaded.
[root@microshift vmi]# oc logs cdi-upload-example-upload-dv -f
I1211 10:03:18.977355 1 uploadserver.go:74] Running server on 0.0.0.0:8443
I1211 10:19:49.234784 1 uploadserver.go:324] Content type header is ""
I1211 10:19:49.237511 1 data-processor.go:379] Calculating available size
I1211 10:19:49.238101 1 data-processor.go:391] Checking out file system volume size.
I1211 10:19:49.238309 1 data-processor.go:399] Request image size not empty.
I1211 10:19:49.238394 1 data-processor.go:404] Target size 10Gi.
I1211 10:19:49.238782 1 data-processor.go:282] New phase: TransferScratch
I1211 10:19:49.242050 1 util.go:191] Writing data...
I1211 10:20:59.306801 1 data-processor.go:282] New phase: ValidatePause
I1211 10:20:59.311513 1 data-processor.go:288] Validating image
I1211 10:20:59.441552 1 data-processor.go:282] New phase: Pause
I1211 10:20:59.441700 1 uploadserver.go:367] Returning success to caller, continue processing in background
I1211 10:20:59.441832 1 data-processor.go:193] Resuming processing at phase Convert
I1211 10:20:59.441913 1 data-processor.go:288] Validating image
I1211 10:23:53.110341 1 data-processor.go:282] New phase: Resize
W1211 10:23:53.176981 1 data-processor.go:361] Available space less than requested size, resizing image to available space 10146021376.
I1211 10:23:53.177109 1 data-processor.go:372] Expanding image size to: 10146021376
I1211 10:23:53.372418 1 data-processor.go:288] Validating image
I1211 10:23:53.409971 1 data-processor.go:282] New phase: Complete
I1211 10:23:53.410194 1 uploadserver.go:364] Wrote data to /data/disk.img
I1211 10:23:53.410273 1 uploadserver.go:203] Shutting down http server after successful upload
I1211 10:23:53.420251 1 uploadserver.go:103] UploadServer successfully exited
The scratch volume and the upload pod will be deleted after the upload is complete. The persistent volume with the CentOS 9 Stream image can now be used as a source to clone multiple VMs. Now you can install the KubeVirt Operator as shown in Sample 5 (if not already installed) and create a VM by cloning this image.
[root@microshift vmi]# oc apply -f vm-centos9-uploadvolume.yaml
virtualmachine.kubevirt.io/centos9instance2 created
[root@microshift vmi]# watch oc get dv,pvc,pods,vm,vmi -n default
NAME PHASE PROGRESS RESTARTS AGE
datavolume.cdi.kubevirt.io/centos9instance2 CloneInProgress N/A 15s
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS
AGE
persistentvolumeclaim/centos9instance2 Bound pvc-3cde8c3f-3d15-42bc-9961-f5c969deb1a4 57Gi RWO kubevirt-hostpath-provisioner
14s
persistentvolumeclaim/example-upload-dv Bound pvc-347eb116-b02f-4790-adb5-adb239e8dcc5 57Gi RWO kubevirt-hostpath-provisioner
23m
NAME READY STATUS RESTARTS AGE
pod/3cde8c3f-3d15-42bc-9961-f5c969deb1a4-source-pod 0/1 ContainerCreating 0 4s
pod/cdi-upload-centos9instance2 1/1 Running 0 10s
NAME AGE STATUS READY
virtualmachine.kubevirt.io/centos9instance2 15s Provisioning False
Update the persistent volume claim annotation kubevirt.io/provisionOnNode to your microshift node name in the vm-centos9-uploadvolume.yaml and apply this yaml. The DataVolume for centos9instance2 will be cloned. You will see the CloneInProgress for the data volume with a percentage of progress. The VM will be provisioned and the VMI for centos9instance2 will be started. You can connect to it as in previous example. When done, run the “oc delete -f vm-centos9-uploadvolume.yaml” to delete the VM and the cloned persistent volume.
You can also use the arm64 Ubuntu Cloud Images. We will run the Ubuntu VM using the https://cloud-images.ubuntu.com/jammy/current/jammy-server-cloudimg-arm64.img. When the VM is started, you can login using virtctl console with userid ubuntu and password ubuntu provided in the vm-ubuntujammy-uploadvolume.yaml. You can also ssh to the VMI ip address (10.42.0.20 shown below).
oc delete example-upload-dv
oc apply -f example-upload-dv.yaml
wget https://cloud-images.ubuntu.com/jammy/current/jammy-server-cloudimg-arm64.img
virtctl image-upload dv example-upload-dv --namespace default --size 10Gi --image-path jammy-server-cloudimg-arm64.img --wait-secs 1200 --no-create --uploadproxy-url=https://cdi-uploadproxy-cdi.cluster.local --insecure
oc apply -f vm-ubuntujammy-uploadvolume.yaml
ssh -i ~/amazontestkey.pem root@10.42.0.20
ssh -i ~/amazontestkey.pem ubuntu@10.42.0.20
You can do the same with the https://cloud-images.ubuntu.com/lunar/current/lunar-server-cloudimg-arm64.img using the vm-ubuntulunar-uploadvolume.yaml.
You can even test by compressing and uploading the lunar-server-cloudimg-arm64.img.xz image.
xz -v lunar-server-cloudimg-arm64.img
Finally, you can delete the CDI resource and Operator as follows:
VERSION=v1.55.2
oc delete -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-cr.yaml
oc delete -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-operator.yaml
Output:
[root@microshift vmi]# oc delete -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-cr.yaml
cdi.cdi.kubevirt.io "cdi" deleted
root@microshift vmi]# oc get pods -n cdi
NAME READY STATUS RESTARTS AGE
cdi-apiserver-7d579fc44c-jgwqs 0/1 Terminating 0 152m
cdi-operator-794d4f98fc-dxv5b 1/1 Running 0 110m
cdi-uploadproxy-6546b64f8f-zp9nv 1/1 Terminating 0 151m
[root@microshift vmi]# oc delete -f https://github.com/kubevirt/containerized-data-importer/releases/download/$VERSION/cdi-operator.yaml
namespace "cdi" deleted
customresourcedefinition.apiextensions.k8s.io "cdis.cdi.kubevirt.io" deleted
clusterrole.rbac.authorization.k8s.io "cdi-operator-cluster" deleted
clusterrolebinding.rbac.authorization.k8s.io "cdi-operator" deleted
serviceaccount "cdi-operator" deleted
role.rbac.authorization.k8s.io "cdi-operator" deleted
rolebinding.rbac.authorization.k8s.io "cdi-operator" deleted
deployment.apps "cdi-operator" deleted
configmap "cdi-operator-leader-election-helper" deleted
7. MongoDB
We will deploy and use the mongodb database using the image: docker.io/arm64v8/mongo:4.4.18. Do not use the latest tag for the image. It will result in "WARNING: MongoDB 5.0+ requires ARMv8.2-A or higher, and your current system does not appear to implement any of the common features for that!" and fail to start. Raspberry Pi 4 uses an ARM Cortex-A72 which is ARM v8-A.
A new PersistentVolumeClaim mongodb will use the storageClassName: kubevirt-hostpath-provisioner for the Persistent Volume. The mongodb-root-username uses the root user with a the mongodb-root-password set to a default of mongodb-password. Update the volume.kubernetes.io/selected-node in mongodb-pv.yaml to your microshift node name.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/mongodb
vi mongodb-pv.yaml # Update the microshift node name
oc project default
oc apply -f .
We create the school db and insert 1000 records into the student collection.
Output:
[root@microshift mongodb]# oc exec -it statefulset/mongodb -- bash
groups: cannot find name for group ID 1001
1001@mongodb-0:/$ mongo admin --host mongodb.default.svc.cluster.local:27017 --authenticationDatabase admin -u root -p mongodb-password
MongoDB shell version v4.4.18
connecting to: mongodb://mongodb.default.svc.cluster.local:27017/admin?authSource=admin&compressors=disabled&gssapiServiceName=mongodb
Implicit session: session { "id" : UUID("604ba70d-da1b-4afa-99d5-00d12f75e3e8") }
MongoDB server version: 4.4.18
Welcome to the MongoDB shell.
For interactive help, type "help".
For more comprehensive documentation, see
https://docs.mongodb.com/
Questions? Try the MongoDB Developer Community Forums
https://community.mongodb.com
---
The server generated these startup warnings when booting:
2022-12-08T17:42:59.611+00:00: Using the XFS filesystem is strongly recommended with the WiredTiger storage engine. See http://dochub.mongodb.org/core/prodnotes-filesystem
---
---
Enable MongoDB's free cloud-based monitoring service, which will then receive and display
metrics about your deployment (disk utilization, CPU, operation statistics, etc).
The monitoring data will be available on a MongoDB website with a unique URL accessible to you
and anyone you share the URL with. MongoDB may use this information to make product
improvements and to suggest MongoDB products and deployment options to you.
To enable free monitoring, run the following command: db.enableFreeMonitoring()
To permanently disable this reminder, run the following command: db.disableFreeMonitoring()
---
> show dbs
admin 0.000GB
config 0.000GB
local 0.000GB
school 0.000GB
> use admin
switched to db admin
> show users
{
"_id" : "admin.root",
"userId" : UUID("658e0dae-b995-4912-aacc-131710f62846"),
"user" : "root",
"db" : "admin",
"roles" : [
{
"role" : "root",
"db" : "admin"
}
],
"mechanisms" : [
"SCRAM-SHA-1",
"SCRAM-SHA-256"
]
}
> use school
switched to db school
> for(var i = 1;i<=1000;i++){ db.students.insert({student_id:i,class:Math.ceil(Math.random()*20),scores:[{type:"exam" , score:Math.ceil(Math.random()*100)},{type:"quiz" , score:Math.ceil(Math.random()*100)},{type:"homework" , score:Math.ceil(Math.random()*100)},{type:"homework" , score:Math.ceil(Math.random()*100)}]}); if(i%100 == 0) {print(i);} }
100
200
300
400
500
600
700
800
900
1000
> db.students.findOne()
{
"_id" : ObjectId("639223deebe39f904788264f"),
"student_id" : 1,
"class" : 19,
"scores" : [
{
"type" : "exam",
"score" : 6
},
{
"type" : "quiz",
"score" : 67
},
{
"type" : "homework",
"score" : 52
},
{
"type" : "homework",
"score" : 11
}
]
}
> exit
bye
1001@mongodb-0:/$ exit
exit
We can run another client pod to connect to the mongodb service and check the number of students in the collection in the school db we created earlier.
[root@microshift mongodb]# oc run --namespace default mongodb-client --rm --tty -i --restart='Never' --image docker.io/arm64v8/mongo:4.4.18 -- bash
If you don't see a command prompt, try pressing enter.
root@mongodb-client:/# mongo admin --host mongodb.default.svc.cluster.local:27017 --authenticationDatabase admin -u root -p mongodb-password
MongoDB shell version v4.4.18
connecting to: mongodb://mongodb.default.svc.cluster.local:27017/admin?authSource=admin&compressors=disabled&gssapiServiceName=mongodb
Implicit session: session { "id" : UUID("d5ed2632-3bec-4527-a8b3-cd7651c14fbb") }
MongoDB server version: 4.4.18
Welcome to the MongoDB shell.
For interactive help, type "help".
For more comprehensive documentation, see
https://docs.mongodb.com/
Questions? Try the MongoDB Developer Community Forums
https://community.mongodb.com
---
The server generated these startup warnings when booting:
2022-12-08T18:10:02.794+00:00: Using the XFS filesystem is strongly recommended with the WiredTiger storage engine. See http://dochub.mongodb.org/core/prodnotes-filesystem
---
---
Enable MongoDB's free cloud-based monitoring service, which will then receive and display
metrics about your deployment (disk utilization, CPU, operation statistics, etc).
The monitoring data will be available on a MongoDB website with a unique URL accessible to you
and anyone you share the URL with. MongoDB may use this information to make product
improvements and to suggest MongoDB products and deployment options to you.
To enable free monitoring, run the following command: db.enableFreeMonitoring()
To permanently disable this reminder, run the following command: db.disableFreeMonitoring()
---
> use school
switched to db school
> db.students.count()
1000
> exit
bye
root@mongodb-client:/# exit
exit
pod "mongodb-client" deleted
When we are done, we can delete the mongodb
cd ~/microshift/raspberry-pi/mongodb
oc delete -f .
You can alternatively install mongodb using the bitnami helm chart, however the auth will not be set because the chart uses different env variables for username and password within the pod.
helm install mongodb bitnami/mongodb --set image.registry=docker.io --set image.repository=arm64v8/mongo --set image.tag=4.4.18 --set persistence.mountPath=/data/db --set livenessProbe.enabled=false --set readinessProbe.enabled=false --set auth.enabled=false
Cleanup MicroShift
We can use the cleanup.sh script available on github to cleanup the pods and images. If you already cloned the microshift repo from github, you have the script in the ~/microshift/hack directory.
cd ~/microshift/hack
./cleanup.sh
Containerized MicroShift on AlmaLinux (64 bit)
We can run MicroShift within containers in two ways:
- MicroShift Containerized – The MicroShift binary runs in a Podman container, CRI-O Systemd service runs directly on the host and data is stored in a podman volume
- MicroShift Containerized All-In-One – The MicroShift binary and CRI-O service run within a container and data is stored in a podman volume, microshift-data. This should be used for “Testing and Development” only
Microshift Containerized
If you did not already install podman, you can do it now.
dnf install -y podman
We will use a new microshift.service that runs microshift in a pod using the prebuilt image and uses a podman volume. Rest of the pods run using crio on the host. We use the image in the microshift.service below with docker.io/karve/microshift:latest that was built using instructions in Part 4. This container image contains the microshift binary that fixes the hugetlb and CDI requestheader-allowed-names errors.
cat << EOF > /usr/lib/systemd/system/microshift.service
[Unit]
Description=MicroShift Containerized
Documentation=man:podman-generate-systemd(1)
Wants=network-online.target crio.service
After=network-online.target crio.service
RequiresMountsFor=%t/containers
[Service]
Environment=PODMAN_SYSTEMD_UNIT=%n
Restart=on-failure
TimeoutSec=3000
ExecStartPre=/usr/bin/mkdir -p /var/lib/kubelet ; /usr/bin/mkdir -p /var/hpvolumes
ExecStartPre=/bin/rm -f %t/%n.ctr-id
ExecStart=/bin/podman run \
--cidfile=%t/%n.ctr-id \
--cgroups=no-conmon \
--rm \
--replace \
--sdnotify=container \
--label io.containers.autoupdate=registry \
--network=host \
--privileged \
-d \
--name microshift \
-v /var/hpvolumes:/var/hpvolumes:z,rw,rshared \
-v /var/run/crio/crio.sock:/var/run/crio/crio.sock:rw,rshared \
-v microshift-data:/var/lib/microshift:rw,rshared \
-v /var/lib/kubelet:/var/lib/kubelet:z,rw,rshared \
-v /var/log:/var/log \
-v /etc:/etc quay.io/microshift/microshift:latest
ExecStop=/bin/podman stop --ignore --cidfile=%t/%n.ctr-id
ExecStopPost=/bin/podman rm -f --ignore --cidfile=%t/%n.ctr-id
Type=notify
NotifyAccess=all
[Install]
WantedBy=multi-user.target default.target
EOF
systemctl daemon-reload
systemctl enable --now crio microshift
podman ps -a
podman volume inspect microshift-data # Get the Mountpoint where kubeconfig is located
export KUBECONFIG=/var/lib/containers/storage/volumes/microshift-data/_data/resources/kubeadmin/kubeconfig
watch "oc get nodes;oc get pods -A;crictl pods;crictl images"
The microshift process may not work, no nodes and no pods, it remains in the following state:
No resources found
No resources found
POD ID CREATED STATE NAME NAMESPACE ATTEMPT RUNTIME
IMAGE TAG IMAGE ID SIZE
quay.io/microshift/microshift latest bdccb7de6c282 406MB
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
31fba8a38d3b quay.io/microshift/microshift:latest run 33 minutes ago Up 33 minutes ago microshift
In above case, the podman logs for microshift show “ResponseCode: 503, Body: service unavailable”.
podman logs -f microshift 2>&1 | grep 503
E1127 16:56:29.039225 1 controller.go:116] loading OpenAPI spec for "v1.user.openshift.io" failed with: failed to retrieve openAPI spec, http error: ResponseCode: 503, Body: service unavailable
Restart the microshift.service and wait for pods to start. I will update this blog if I find the cause and a better way to fix this!
[root@microshift hack]# podman exec -it microshift kill 1
[root@microshift hack]# systemctl start microshift
or
[root@microshift hack]# systemctl restart microshift
You will see errors about the hugepages with default image because the /proc/cgroups does not have hugepages. You may ignore them. Alternatively, this can be fixed by rebuilding the microshift binary as shown in later section (docker.io/karve/microshift:latest) and updating the microshift container.
W1127 17:45:34.302111 1 container.go:586] Failed to update stats for container "/system.slice": error while statting cgroup v2: [open /sys/kernel/mm/hugepages: no such file or directory
failed to fetch hugetlb info
github.com/opencontainers/runc/libcontainer/cgroups/fs2.statHugeTlb
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs2/hugetlb.go:35
github.com/opencontainers/runc/libcontainer/cgroups/fs2.(*manager).GetStats
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs2/fs2.go:123
github.com/google/cadvisor/container/libcontainer.(*Handler).GetStats
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/google/cadvisor/container/libcontainer/handler.go:83
github.com/google/cadvisor/container/raw.(*rawContainerHandler).GetStats
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/google/cadvisor/container/raw/handler.go:232
github.com/google/cadvisor/manager.(*containerData).updateStats
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/google/cadvisor/manager/container.go:637
github.com/google/cadvisor/manager.(*containerData).housekeepingTick
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/google/cadvisor/manager/container.go:583
github.com/google/cadvisor/manager.(*containerData).housekeeping
/opt/app-root/src/github.com/redhat-et/microshift/vendor/github.com/google/cadvisor/manager/container.go:531
runtime.goexit
/usr/lib/golang/src/runtime/asm_arm64.s:1130], continuing to push stats
Now that microshift is started, we can run the samples shown earlier.
After we are done, we can delete the microshift container. The --rm we used in the podman run will delete the container when we stop it.
podman stop microshift && podman volume rm microshift-data
After it is stopped, we can run the cleanup.sh to delete the pods and images from crio.
MicroShift Containerized All-In-One
Let’s stop the crio on the host, we will be creating an all-in-one container in podman that will run crio within the container.
systemctl stop crio
systemctl disable crio
mkdir /var/hpvolumes
We will run the all-in-one microshift in podman using prebuilt images (replace the image in the podman run command below with the latest image). Normally you would run the following to start the all-in-one microshift, but it does not work. You can try to restart microshift with “systemctl restart microshift”, that won’t get the node to the Ready state.
setsebool -P container_manage_cgroup true
podman volume rm microshift-data;podman volume create microshift-data
podman run -d --rm --name microshift -h microshift-aio.example.com --privileged -v /lib/modules:/lib/modules -v microshift-data:/var/lib -v /var/hpvolumes:/var/hpvolumes -p 6443:6443 -p 8080:8080 -p 80:80 quay.io/microshift/microshift-aio:4.8.0-0.microshift-2022-04-20-182108-linux-nft-arm64
The microshift logs within the container will show Error in setting cgroup config for procHooks
Nov 27 18:32:46 microshift-aio.example.com microshift[731]: E1127 18:32:46.095391 731 pod_workers.go:190] "Error syncing pod, skipping" err="failed to \"CreatePodSandbox\" for \"kube-flannel-ds-65q9m_kube-system(9f02c0dc-540b-42c1-89ff-ad545b30131d)\" with CreatePodSandboxError: \"Failed to create sandbox for pod \\\"kube-flannel-ds-65q9m_kube-system(9f02c0dc-540b-42c1-89ff-ad545b30131d)\\\": rpc error: code = Unknown desc = container create failed: time=\\\"2022-11-27T18:32:45Z\\\" level=warning msg=\\\"unable to get oom kill count\\\" error=\\\"openat2 /sys/fs/cgroup/system.slice/runc-d265db5497b3bfec60cd9f0b9254aa103973df2edcfade304692c663e1e1a958.scope/memory.events: no such file or directory\\\"\\ntime=\\\"2022-11-27T18:32:45Z\\\" level=error msg=\\\"container_linux.go:380: starting container process caused: process_linux.go:545: container init caused: process_linux.go:508: setting cgroup config for procHooks process caused: openat2 /sys/fs/cgroup/system.slice/runc-d265db5497b3bfec60cd9f0b9254aa103973df2edcfade304692c663e1e1a958.scope/cpu.weight: no such file or directory\\\"\\n\"" pod="kube-system/kube-flannel-ds-65q9m" podUID=9f02c0dc-540b-42c1-89ff-ad545b30131d
Nov 27 18:32:47 microshift-aio.example.com microshift[731]: E1127 18:32:47.463863 731 pod_workers.go:190] "Error syncing pod, skipping" err="network is not ready: container runtime network not ready: NetworkReady=false reason:NetworkPluginNotReady message:Network plugin returns error: No CNI configuration file in /etc/cni/net.d/. Has your network provider started?" pod="openshift-dns/dns-default-8qrnk" podUID=5f6607e3-f280-4f27-8e29-d059a42a1911
Nov 27 18:32:48 microshift-aio.example.com microshift[731]: E1127 18:32:48.173648 731 kubelet.go:2223] "Container runtime network not ready" networkReady="NetworkReady=false reason:NetworkPluginNotReady message:Network plugin returns error: No CNI configuration file in /etc/cni/net.d/. Has your network provider started?"
Let’s fix this. You can stop the above non-working microshift container with
podman stop microshift
Since the “setsebool -P container_manage_cgroup true” does not work, we mount the /sys/fs/cgroup into the container using -v /sys/fs/cgroup:/sys/fs/cgroup:ro. This will volume mount /sys/fs/cgroup into the container as read/only, but the subdir/mount points will be mounted in as read/write.
podman run -d --rm --name microshift -h microshift-aio.example.com --privileged -v /sys/fs/cgroup:/sys/fs/cgroup:ro -v /lib/modules:/lib/modules -v microshift-data:/var/lib -v /var/hpvolumes:/var/hpvolumes -p 6443:6443 -p 8080:8080 -p 80:80 quay.io/microshift/microshift-aio:4.8.0-0.microshift-2022-04-20-182108-linux-nft-arm64
You will see the hugepages error in the microshift logs that you can ignore (or recompile the microshift binary as shown later and use it within the container)
Nov 27 18:38:36 microshift-aio.example.com microshift[67]: W1127 18:38:36.566697 67 container.go:586] Failed to update stats for container "/system.slice/crio-4d14e2f74d53bab169368fd9fc1295a8b24ffb0f7e17c0a781cef8eaf6c89d3b.scope": error while statting cgroup v2: [open /sys/kernel/mm/hugepages: no such file or directory
Also, you will initially see the flannel errors “open /run/flannel/subnet.env: no such file or directory”. You can ignore these. These errors will resolve after the flannel pod starts and creates the subnet.env.
Nov 27 19:36:55 microshift-aio.example.com microshift[69]: E1127 19:36:55.469267 69 pod_workers.go:190] "Error syncing pod, skipping" err="failed to \"CreatePodSandbox\" for \"kubevirt-hostpath-provisioner-vcdmn_kubevirt-hostpath-provisioner(3b99266e-2387-406e-ad67-d6839807b9d3)\" with CreatePodSandboxError: \"Failed to create sandbox for pod \\\"kubevirt-hostpath-provisioner-vcdmn_kubevirt-hostpath-provisioner(3b99266e-2387-406e-ad67-d6839807b9d3)\\\": rpc error: code = Unknown desc = failed to create pod network sandbox k8s_kubevirt-hostpath-provisioner-vcdmn_kubevirt-hostpath-provisioner_3b99266e-2387-406e-ad67-d6839807b9d3_0(1f1c02ab07d59ad32c530747e653b0c6171063022d0275df534d3bf69a0ac553): error adding pod kubevirt-hostpath-provisioner_kubevirt-hostpath-provisioner-vcdmn to CNI network \\\"cbr0\\\": open /run/flannel/subnet.env: no such file or directory\"" pod="kubevirt-hostpath-provisioner/kubevirt-hostpath-provisioner-vcdmn" podUID=3b99266e-2387-406e-ad67-d6839807b9d3
Now that you know the podman command to start the microshift all-in-one, you may alternatively use the following microshift service.
cat << EOF > /usr/lib/systemd/system/microshift.service
[Unit]
Description=MicroShift all-in-one
Documentation=man:podman-generate-systemd(1)
Wants=network-online.target
After=network-online.target
RequiresMountsFor=%t/containers
[Service]
Environment=PODMAN_SYSTEMD_UNIT=%n
Restart=on-failure
TimeoutSec=3000
ExecStartPre=/bin/rm -f %t/%n.ctr-id
ExecStart=/usr/bin/podman run --cidfile=%t/%n.ctr-id --sdnotify=conmon --cgroups=no-conmon --rm --replace -d --name microshift -h microshift-aio.example.com --privileged -v /sys/fs/cgroup:/sys/fs/cgroup:ro -v microshift-data:/var/lib -v /var/hpvolumes:/var/hpvolumes -v /lib/modules:/lib/modules --label io.containers.autoupdate=registry -p 6443:6443 -p 80:80 quay.io/microshift/microshift-aio:latest
ExecStop=/usr/bin/podman stop --ignore --cidfile=%t/%n.ctr-id
ExecStopPost=/usr/bin/podman rm -f --ignore --cidfile=%t/%n.ctr-id
Type=notify
NotifyAccess=all
[Install]
WantedBy=multi-user.target default.target
EOF
systemctl daemon-reload
systemctl start microshift
Note that the “-h microshift-aio.example.com” hostname used in the container should be different from the hostname of the Raspberry Pi (microshift.example.com). On the host Raspberry Pi 4, we set KUBECONFIG to point to the kubeconfig on the data volume.
export KUBECONFIG=/var/lib/containers/storage/volumes/microshift-data/_data/microshift/resources/kubeadmin/kubeconfig
# crio on host is stopped, so we do not run crictl commands on host
watch "oc get nodes;oc get pods -A;podman exec -it microshift crictl ps -a"
The crio service is stopped on the Raspberry Pi, so crictl command will not work directly on the Pi. The crictl commands will work within the microshift container in podman as shown in the watch command above. You will need to wait for the new pods to be started as seen with crictl (the oc pods command may still show the previous pods from etcd).
Now, we can run the samples shown earlier. To run the Virtual Machine examples in the all-in-one MicroShift, we need to execute the mount with --make-shared as follows in the microshift container to prevent the “Error: path "/var/run/kubevirt" is mounted on "/" but it is not a shared mount” event from virt-handler.
podman exec -it microshift mount --make-shared /
We may also preload the virtual machine images using "crictl pull".
podman exec -it microshift crictl pull quay.io/kubevirt/fedora-cloud-container-disk-demo:20210811_9fec1f849-arm64
For the Virtual Machine Instance Sample 4, we can connect to the vmi-fedora by exposing the ssh port for the Virtual Machine Instance as a NodePort Service after the instance is started. This NodePort is within the all-in-one pod that is running in podman. The ip address of the all-in-one microshift podman container is 10.88.0.6. We expose the target port 22 on the VM as a service on port 22 that is in turn exposed on the microshift container with allocated port 32574 as seen below. We run and exec into a new pod called ssh-proxy, install the openssh-client on the ssh-proxy and ssh to the port 32574 on the all-in-one microshift container. This takes us to the VMI port 22 as shown below:
[root@microshift vmi]# oc get vmi,pods
NAME AGE PHASE IP NODENAME READY
virtualmachineinstance.kubevirt.io/vmi-fedora 7m55s Running 10.42.0.16 microshift-aio.example.com True
NAME READY STATUS RESTARTS AGE
pod/virt-launcher-vmi-fedora-k9bbt 2/2 Running 0 6m41s
[root@microshift vmi]# virtctl expose vmi vmi-fedora --port=22 --target-port=22 --name=vmi-fedora-ssh --type=NodePort
Service vmi-fedora-ssh successfully exposed for vmi vmi-fedora
[root@microshift vmi]# oc get svc vmi-fedora-ssh
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
vmi-fedora-ssh NodePort 10.43.181.142 <none> 22:32574/TCP 9s
[root@microshift vmi]# podman inspect --format "{{.NetworkSettings.IPAddress}}" microshift
10.88.0.6
[root@microshift vmi]# oc run -i --tty ssh-proxy --rm --image=karve/alpine-sshclient:arm64 --restart=Never -- /bin/sh -c "ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null fedora@10.88.0.6 -p 32574"
If you don't see a command prompt, try pressing enter.
[fedora@vmi-fedora ~]$ sudo dnf install -y qemu-guest-agent >/dev/null
[fedora@vmi-fedora ~]$ sudo systemctl enable --now qemu-guest-agent
[fedora@vmi-fedora ~]$ ping -c 2 google.com
PING google.com (142.250.72.110) 56(84) bytes of data.
64 bytes from lga34s32-in-f14.1e100.net (142.250.72.110): icmp_seq=1 ttl=115 time=5.88 ms
64 bytes from lga34s32-in-f14.1e100.net (142.250.72.110): icmp_seq=2 ttl=115 time=5.13 ms
--- google.com ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 5.132/5.505/5.878/0.373 ms
[fedora@vmi-fedora ~]$ exit
logout
Connection to 10.88.0.6 closed.
pod "ssh-proxy" deleted
We can install the QEMU guest agent daemon that runs on the virtual machine and passes information to the host about the virtual machine, users, file systems, and secondary networks.
After we are done, we can delete the all-in-one microshift container.
podman rm -f microshift && podman volume rm microshift-data
or if started using systemd, then
systemctl stop microshift
Build and install the Kata Containers
Run the cleanup shown earlier, then create the service for running MicroShift non-containerized.
cat << EOF > /usr/lib/systemd/system/microshift.service
[Unit]
Description=MicroShift
Wants=network-online.target crio.service
After=network-online.target crio.service
[Service]
WorkingDirectory=/usr/bin/
ExecStart=microshift run
Restart=always
User=root
[Install]
WantedBy=multi-user.target
EOF
systemctl daemon-reload
Build and install kata-runtime
go get -d -u github.com/kata-containers/kata-containers
cd /root/go/src/github.com/kata-containers/kata-containers/src/runtime/
make
make install
Output:
[root@microshift runtime]# make
GENERATE data/kata-collect-data.sh
GENERATE pkg/katautils/config-settings.go
kata-runtime - version 3.1.0-alpha0 (commit 9bde32daa102368b9dbc27a6c03ed2e3e87d65e1)
• architecture:
Host:
golang:
Build: arm64
• golang:
go version go1.19.3 linux/arm64
• hypervisors:
Default: qemu
Known: acrn cloud-hypervisor firecracker qemu
Available for this architecture: cloud-hypervisor firecracker qemu
• Summary:
destination install path (DESTDIR) : /
binary installation path (BINDIR) : /usr/local/bin
binaries to install :
- /usr/local/bin/kata-runtime
- /usr/local/bin/containerd-shim-kata-v2
- /usr/local/bin/kata-monitor
- /usr/local/bin/data/kata-collect-data.sh
configs to install (CONFIGS) :
- config/configuration-clh.toml
- config/configuration-fc.toml
- config/configuration-qemu.toml
install paths (CONFIG_PATHS) :
- /usr/share/defaults/kata-containers/configuration-clh.toml
- /usr/share/defaults/kata-containers/configuration-fc.toml
- /usr/share/defaults/kata-containers/configuration-qemu.toml
alternate config paths (SYSCONFIG_PATHS) :
- /etc/kata-containers/configuration-clh.toml
- /etc/kata-containers/configuration-fc.toml
- /etc/kata-containers/configuration-qemu.toml
default install path for qemu (CONFIG_PATH) : /usr/share/defaults/kata-containers/configuration.toml
default alternate config path (SYSCONFIG) : /etc/kata-containers/configuration.toml
qemu hypervisor path (QEMUPATH) : /usr/bin/qemu-system-aarch64
cloud-hypervisor hypervisor path (CLHPATH) : /usr/bin/cloud-hypervisor
firecracker hypervisor path (FCPATH) : /usr/bin/firecracker
assets path (PKGDATADIR) : /usr/share/kata-containers
shim path (PKGLIBEXECDIR) : /usr/libexec/kata-containers
BUILD /root/go/src/github.com/kata-containers/kata-containers/src/runtime/kata-runtime
GENERATE config/configuration-qemu.toml
GENERATE config/configuration-clh.toml
GENERATE config/configuration-fc.toml
BUILD /root/go/src/github.com/kata-containers/kata-containers/src/runtime/containerd-shim-kata-v2
BUILD /root/go/src/github.com/kata-containers/kata-containers/src/runtime/kata-monitor
[root@microshift runtime]# make install
kata-runtime - version 3.1.0-alpha0 (commit 9bde32daa102368b9dbc27a6c03ed2e3e87d65e1)
• architecture:
Host:
golang:
Build: arm64
• golang:
go version go1.19.3 linux/arm64
• hypervisors:
Default: qemu
Known: acrn cloud-hypervisor firecracker qemu
Available for this architecture: cloud-hypervisor firecracker qemu
• Summary:
destination install path (DESTDIR) : /
binary installation path (BINDIR) : /usr/local/bin
binaries to install :
- /usr/local/bin/kata-runtime
- /usr/local/bin/containerd-shim-kata-v2
- /usr/local/bin/kata-monitor
- /usr/local/bin/data/kata-collect-data.sh
configs to install (CONFIGS) :
- config/configuration-clh.toml
- config/configuration-fc.toml
- config/configuration-qemu.toml
install paths (CONFIG_PATHS) :
- /usr/share/defaults/kata-containers/configuration-clh.toml
- /usr/share/defaults/kata-containers/configuration-fc.toml
- /usr/share/defaults/kata-containers/configuration-qemu.toml
alternate config paths (SYSCONFIG_PATHS) :
- /etc/kata-containers/configuration-clh.toml
- /etc/kata-containers/configuration-fc.toml
- /etc/kata-containers/configuration-qemu.toml
default install path for qemu (CONFIG_PATH) : /usr/share/defaults/kata-containers/configuration.toml
default alternate config path (SYSCONFIG) : /etc/kata-containers/configuration.toml
qemu hypervisor path (QEMUPATH) : /usr/bin/qemu-system-aarch64
cloud-hypervisor hypervisor path (CLHPATH) : /usr/bin/cloud-hypervisor
firecracker hypervisor path (FCPATH) : /usr/bin/firecracker
assets path (PKGDATADIR) : /usr/share/kata-containers
shim path (PKGLIBEXECDIR) : /usr/libexec/kata-containers
INSTALL install-scripts
INSTALL install-completions
INSTALL install-configs
INSTALL install-configs
INSTALL install-bin
INSTALL install-containerd-shim-v2
INSTALL install-monitor
The build creates the following:
- runtime binary: /usr/local/bin/kata-runtime and /usr/local/bin/containerd-shim-kata-v2
- configuration file: /usr/share/defaults/kata-containers/configuration.toml
Output:
[root@microshift runtime]# ls -las /usr/local/bin/kata-runtime /usr/local/bin/containerd-shim-kata-v2 /usr/share/defaults/kata-containers/configuration.toml
36588 -rwxr-xr-x. 1 root root 37459871 Nov 24 10:55 /usr/local/bin/containerd-shim-kata-v2
36164 -rwxr-xr-x. 1 root root 37063056 Nov 24 10:55 /usr/local/bin/kata-runtime
0 lrwxrwxrwx. 1 root root 23 Nov 24 10:55 /usr/share/defaults/kata-containers/configuration.toml -> configuration-qemu.toml
Check hardware requirements
kata-runtime check --verbose # This will return error because vmlinux.container does not exist yet
which kata-runtime
kata-runtime --version
containerd-shim-kata-v2 --version
Output:
[root@microshift runtime]# kata-runtime check --verbose
ERRO[0000] /usr/share/defaults/kata-containers/configuration-qemu.toml: file /usr/bin/qemu-system-aarch64 does not exist arch=arm64 name=kata-runtime pid=19912 source=runtime
/usr/share/defaults/kata-containers/configuration-qemu.toml: file /usr/bin/qemu-system-aarch64 does not exist
[root@microshift runtime]# which kata-runtime
/usr/local/bin/kata-runtime
[root@microshift runtime]# kata-runtime --version
kata-runtime : 3.1.0-alpha0
commit : 9bde32daa102368b9dbc27a6c03ed2e3e87d65e1
OCI specs: 1.0.2-dev
[root@microshift runtime]# containerd-shim-kata-v2 --version
Kata Containers containerd shim: id: "io.containerd.kata.v2", version: 3.1.0-alpha0, commit: 9bde32daa102368b9dbc27a6c03ed2e3e87d65e1
Kata creates a VM in which to run one or more containers by launching a hypervisor. Kata supports multiple hypervisors. We use QEMU. The hypervisor needs two assets for this task: a Linux kernel and a small root filesystem image to boot the VM. The guest kernel is passed to the hypervisor and used to boot the VM. The default kernel provided in Kata Containers is highly optimized for kernel boot time and minimal memory footprint, providing only those services required by a container workload. The hypervisor uses an image file which provides a minimal root filesystem used by the guest kernel to boot the VM and host the Kata Container. Kata Containers supports both initrd and rootfs based minimal guest images. The default packages provide both an image and an initrd, both of which are created using the osbuilder tool.
The initrd image is a compressed cpio(1) archive, created from a rootfs which is loaded into memory and used as part of the Linux startup process. During startup, the kernel unpacks it into a special instance of a tmpfs mount that becomes the initial root filesystem.
- The runtime will launch the configured hypervisor.
- The hypervisor will boot the mini-OS image using the guest kernel.
- The kernel will start the init daemon as PID 1 (the agent) inside the VM root environment.
- The agent will create a new container environment, setting its root filesystem to that requested by the user.
- The agent will then execute the command inside the new container.
The rootfs image, sometimes referred to as the mini O/S, is a highly optimized container bootstrap system. With this,
- The runtime will launch the configured hypervisor.
- The hypervisor will boot the mini-OS image using the guest kernel.
- The kernel will start the init daemon as PID 1 (systemd) inside the VM root environment.
- systemd, running inside the mini-OS context, will launch the agent in the root context of the VM.
- The agent will create a new container environment, setting its root filesystem to that requested by the user.
- The agent will then execute the command inside the new container.
We will see the output of the qemu commands later.
Configure to use initrd image
Since, Kata containers can run with either an initrd image or a rootfs image, we will build both images but initially use the initrd. We will switch to rootfs in later section. So, make sure you add initrd = /usr/share/kata-containers/kata-containers-initrd.img in the configuration file /usr/share/defaults/kata-containers/configuration.toml and comment out the default image line with the following:
sudo mkdir -p /etc/kata-containers/
sudo install -o root -g root -m 0640 /usr/share/defaults/kata-containers/configuration.toml /etc/kata-containers
sudo sed -i 's/^\(image =.*\)/# \1/g' /etc/kata-containers/configuration.toml
The initrd line is not added by default, so add the initrd line in /etc/kata-containers/configuration.toml so that it looks as follows:
# image = "/usr/share/kata-containers/kata-containers.img"
initrd = "/usr/share/kata-containers/kata-containers-initrd.img"
Next, we create the initrd image and the rootfs image. One of the initrd and image options in Kata runtime config file must be set, but not both. The main difference between the options is that the size of initrd (10MB+) is significantly smaller than rootfs image (100MB+).
Create an initrd image
Create a local rootfs for initrd image
yum -y install podman buildah skopeo # If not already installed
sudo ln -s `which podman` /usr/bin/docker # or run later commands with USE_PODMAN instead of USE_DOCKER
export ROOTFS_DIR="${GOPATH}/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder/rootfs"
sudo rm -rf ${ROOTFS_DIR}
cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder
./rootfs.sh -l
Output:
[root@microshift runtime]# cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder
[root@microshift rootfs-builder]# ./rootfs.sh -l
alpine
centos
clearlinux
debian
ubuntu
Let’s use the distro=ubuntu
export distro=ubuntu
script -fec 'sudo -E GOPATH=$GOPATH AGENT_INIT=yes USE_PODMAN=true ./rootfs.sh ${distro}'
This will install rust in the container, download and compile numerous crates using cargo and build the kata-agent (in Rust). When complete, you will see the rootfs directory.
[root@microshift rootfs-builder]# ls rootfs
bin boot dev etc home lib lib64 media mnt opt proc root run sbin srv sys tmp usr var
If you get errors such as following, just run the above command again after some time.
Could not connect to ports.ubuntu.com:80 (185.125.190.39), connection timed out
Build an initrd image
cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/initrd-builder
script -fec 'sudo -E AGENT_INIT=yes USE_PODMAN=true ./initrd_builder.sh ${ROOTFS_DIR}'
Output:
[root@microshift rootfs-builder]# cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/initrd-builder
[root@microshift initrd-builder]# script -fec 'sudo -E AGENT_INIT=yes USE_PODMAN=true ./initrd_builder.sh ${ROOTFS_DIR}'
Script started, output log file is 'typescript'.
[OK] init is installed
[OK] Agent is installed
INFO: Creating /root/go/src/github.com/kata-containers/kata-containers/tools/osbuilder/initrd-builder/kata-containers-initrd.img based on rootfs at /root/go/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder/rootfs
136212 blocks
Script done.
Install the initrd image
commit=$(git log --format=%h -1 HEAD)
date=$(date +%Y-%m-%d-%T.%N%z)
image="kata-containers-initrd-${date}-${commit}"
sudo install -o root -g root -m 0640 -D kata-containers-initrd.img "/usr/share/kata-containers/${image}"
(cd /usr/share/kata-containers && sudo ln -sf "$image" kata-containers-initrd.img)
Output:
[root@microshift initrd-builder]# commit=$(git log --format=%h -1 HEAD)
[root@microshift initrd-builder]# date=$(date +%Y-%m-%d-%T.%N%z)
[root@microshift initrd-builder]# image="kata-containers-initrd-${date}-${commit}"
[root@microshift initrd-builder]# sudo install -o root -g root -m 0640 -D kata-containers-initrd.img "/usr/share/kata-containers/${image}"
[root@microshift initrd-builder]# (cd /usr/share/kata-containers && sudo ln -sf "$image" kata-containers-initrd.img)
Create a rootfs image
Create a local rootfs for rootfs image
export ROOTFS_DIR=${GOPATH}/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder/rootfs
sudo rm -rf ${ROOTFS_DIR}
cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/rootfs-builder
script -fec 'sudo -E GOPATH=$GOPATH USE_PODMAN=true ./rootfs.sh ${distro}'
Build a rootfs image
cd $GOPATH/src/github.com/kata-containers/kata-containers/tools/osbuilder/image-builder
script -fec 'sudo -E USE_PODMAN=true ./image_builder.sh ${ROOTFS_DIR}'
Install the rootfs image
commit=$(git log --format=%h -1 HEAD)
date=$(date +%Y-%m-%d-%T.%N%z)
image="kata-containers-${date}-${commit}"
sudo install -o root -g root -m 0640 -D kata-containers.img "/usr/share/kata-containers/${image}"
(cd /usr/share/kata-containers && sudo ln -sf "$image" kata-containers.img)
Build Kata Containers Kernel
The process to build a kernel for Kata Containers is automated.
yum -y install flex bison bc
go env -w GO111MODULE=auto
go get github.com/kata-containers/packaging
cd $GOPATH/src/github.com/kata-containers/packaging/kernel
The script ./build-kernel.sh tries to apply the patches from ${GOPATH}/src/github.com/kata-containers/packaging/kernel/patches/ when it sets up a kernel. If you want to add a source modification, add a patch on this directory. The script also copies or generates a kernel config file from ${GOPATH}/src/github.com/kata-containers/packaging/kernel/configs/ to .config in the kernel source code. You can modify it as needed. We will use the defaults.
./build-kernel.sh setup
Output:
[root@microshift image-builder]# go env -w GO111MODULE=auto
[root@microshift image-builder]# go get github.com/kata-containers/packaging
package github.com/kata-containers/packaging: no Go files in /root/go/src/github.com/kata-containers/packaging
[root@microshift image-builder]# cd $GOPATH/src/github.com/kata-containers/packaging/kernel
[root@microshift kernel]# ./build-kernel.sh setup
package github.com/kata-containers/tests: no Go files in /root/go/src/github.com/kata-containers/tests
~/go/src/github.com/kata-containers/tests ~/go/src/github.com/kata-containers/packaging/kernel
~/go/src/github.com/kata-containers/packaging/kernel
versions file (versions-master.yaml) does not exist
Download from https://raw.githubusercontent.com/kata-containers/runtime/master/versions.yaml
INFO: Config version: 92
INFO: Kernel version: 5.4.60
INFO: kernel path does not exist, will download kernel
INFO: Download kernel checksum file: sha256sums.asc
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 266k 100 266k 0 0 321k 0 --:--:-- --:--:-- --:--:-- 321k
INFO: Download kernel version 5.4.60
INFO: Download kernel
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 162 100 162 0 0 207 0 --:--:-- --:--:-- --:--:-- 207
100 104M 100 104M 0 0 7745k 0 0:00:13 0:00:13 --:--:-- 9237k
linux-5.4.60.tar.xz: OK
INFO: Apply patches from /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x
INFO: Found 7 patches
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0001-9p-retrieve-fid-from-file-when-file-instance-exist.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0001-NO-UPSTREAM-9P-always-use-cached-inode-to-fill-in-v9.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0002-ACPI-Always-build-evged-in.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0002-net-virtio_vsock-Fix-race-condition-between-bind-and.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0003-arm-arm64-Provide-a-wrapper-for-SMCCC-1.1-calls.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0004-arm-arm64-smccc-psci-add-arm_smccc_1_1_get_conduit.patch
INFO: Apply /root/go/src/github.com/kata-containers/packaging/kernel/patches//5.4.x/0006-arm64-mm-Enable-memory-hot-remove.patch
ls: cannot access '/root/go/src/github.com/kata-containers/packaging/kernel/configs/fragments/arm64/../common/experimental/*.conf': No such file or directory
INFO: Constructing config from fragments: /root/go/src/github.com/kata-containers/packaging/kernel/configs/fragments/arm64/.config
INFO: Some CONFIG elements failed to make the final .config:
INFO: Value requested for CONFIG_MEMORY_HOTREMOVE not in final .config
Value requested for CONFIG_ZONE_DEVICE not in final .config
Value requested for CONFIG_DEV_PAGEMAP_OPS not in final .config
Value requested for CONFIG_ND_PFN not in final .config
Value requested for CONFIG_NVDIMM_PFN not in final .config
Value requested for CONFIG_NVDIMM_DAX not in final .config
INFO: Generated config file can be found in /root/go/src/github.com/kata-containers/packaging/kernel/configs/fragments/arm64/.config
ERROR: Failed to construct requested .config file
ERROR: failed to find default config
After the kernel source code is ready, we build the kernel
cp /root/go/src/github.com/kata-containers/packaging/kernel/configs/fragments/arm64/.config kata-linux-5.4.60-92/.config
./build-kernel.sh build
Output:
[root@microshift kernel]# cp /root/go/src/github.com/kata-containers/packaging/kernel/configs/fragments/arm64/.config kata-linux-5.4.60-92/.config
[root@microshift kernel]# cd kata-linux-5.4.60-92
[root@microshift kata-linux-5.4.60-92]# make oldconfig
scripts/kconfig/conf --oldconfig Kconfig
#
# No change to .config
#
[root@microshift kata-linux-5.4.60-92]# cd ..
[root@microshift kernel]# ./build-kernel.sh build
…
Install the kernel to the default Kata containers path (/usr/share/kata-containers/)
./build-kernel.sh install
Output:
[root@microshift kernel]# ./build-kernel.sh install
package github.com/kata-containers/tests: no Go files in /root/go/src/github.com/kata-containers/tests
~/go/src/github.com/kata-containers/tests ~/go/src/github.com/kata-containers/packaging/kernel
~/go/src/github.com/kata-containers/packaging/kernel
INFO: Config version: 92
INFO: Kernel version: 5.4.60
CALL scripts/atomic/check-atomics.sh
CALL scripts/checksyscalls.sh
CHK include/generated/compile.h
lrwxrwxrwx. 1 root root 17 Nov 24 15:17 /usr/share/kata-containers/vmlinux.container -> vmlinux-5.4.60-92
lrwxrwxrwx. 1 root root 17 Nov 24 15:17 /usr/share/kata-containers/vmlinuz.container -> vmlinuz-5.4.60-92
The /etc/kata-containers/configuration.toml has the following:
# Path to vhost-user-fs daemon.
virtio_fs_daemon = "/usr/libexec/virtiofsd"
Check the output kata-runtime:
[root@microshift kernel]# kata-runtime check --verbose
ERRO[0000] /etc/kata-containers/configuration.toml: file /usr/bin/qemu-system-aarch64 does not exist arch=arm64 name=kata-runtime pid=205921 source=runtime
/etc/kata-containers/configuration.toml: file /usr/bin/qemu-system-aarch64 does not exist
Let’s fix this with:
ln -s /usr/libexec/qemu-kvm /usr/bin/qemu-system-aarch64
Output:
[root@microshift kernel]# ln -s /usr/libexec/qemu-kvm /usr/bin/qemu-system-aarch64
[root@microshift kernel]# kata-runtime check --verbose
WARN[0000] Not running network checks as super user arch=arm64 name=kata-runtime pid=206229 source=runtime
INFO[0000] Unable to know if the system is running inside a VM arch=arm64 source=virtcontainers/hypervisor
INFO[0000] kernel property found arch=arm64 description="Kernel-based Virtual Machine" name=kvm pid=206229 source=runtime type=module
INFO[0000] kernel property found arch=arm64 description="Host kernel accelerator for virtio" name=vhost pid=206229 source=runtime type=module
INFO[0000] kernel property found arch=arm64 description="Host kernel accelerator for virtio network" name=vhost_net pid=206229 source=runtime type=module
INFO[0000] kernel property found arch=arm64 description="Host Support for Linux VM Sockets" name=vhost_vsock pid=206229 source=runtime type=module
System is capable of running Kata Containers
INFO[0000] device available arch=arm64 check-type=full device=/dev/kvm name=kata-runtime pid=206229 source=runtime
INFO[0000] feature available arch=arm64 check-type=full feature=create-vm name=kata-runtime pid=206229 source=runtime
INFO[0000] kvm extension is supported arch=arm64 description="Maximum IPA shift supported by the host" id=165 name=KVM_CAP_ARM_VM_IPA_SIZE pid=206229 source=runtime type="kvm extension"
INFO[0000] IPA limit size: 44 bits. arch=arm64 name=KVM_CAP_ARM_VM_IPA_SIZE pid=206229 source=runtime type="kvm extension"
System can currently create Kata Containers
Check the hypervisor.qemu section in configuration.toml:
[root@microshift kernel]# cat /etc/kata-containers/configuration.toml | awk -v RS= '/\[hypervisor.qemu\]/'
[hypervisor.qemu]
path = "/usr/bin/qemu-system-aarch64"
kernel = "/usr/share/kata-containers/vmlinux.container"
# image = "/usr/share/kata-containers/kata-containers.img"
initrd = "/usr/share/kata-containers/kata-containers-initrd.img"
machine_type = "virt"
Check the initrd image (kata-containers-initrd.img), the rootfs image (kata-containers.img), and the kernel in the /usr/share/kata-containers directory:
[root@microshift kernel]# ls -las /usr/share/kata-containers
total 171696
4 drwxr-xr-x. 2 root root 4096 Nov 24 15:17 .
4 drwxr-xr-x. 128 root root 4096 Nov 24 14:20 ..
68 -rw-r--r--. 1 root root 68536 Nov 24 15:17 config-5.4.60
131072 -rw-r-----. 1 root root 134217728 Nov 24 14:20 kata-containers-2022-11-24-14:19:58.932179446+0000-9bde32daa
26144 -rw-r-----. 1 root root 26770481 Nov 24 12:48 kata-containers-initrd-2022-11-24-12:48:10.440333043+0000-9bde32daa
4 lrwxrwxrwx. 1 root root 67 Nov 24 12:48 kata-containers-initrd.img -> kata-containers-initrd-2022-11-24-12:48:10.440333043+0000-9bde32daa
4 lrwxrwxrwx. 1 root root 60 Nov 24 14:20 kata-containers.img -> kata-containers-2022-11-24-14:19:58.932179446+0000-9bde32daa
9820 -rw-r--r--. 1 root root 10246656 Nov 24 15:17 vmlinux-5.4.60-92
0 lrwxrwxrwx. 1 root root 17 Nov 24 15:17 vmlinux.container -> vmlinux-5.4.60-92
4576 -rw-r--r--. 1 root root 4684131 Nov 24 15:17 vmlinuz-5.4.60-92
0 lrwxrwxrwx. 1 root root 17 Nov 24 15:17 vmlinuz.container -> vmlinuz-5.4.60-92
The kernel file is called vmlinuz-version. vmlinuz is the name of the Linux kernel executable. vmlinuz is a compressed Linux kernel, and it can load the operating system into memory so that the computer becomes usable and application programs can be run. When virtual memory was developed for easier multitasking abilities, “vm” was put at the front of the file to show that the kernel supports virtual memory. For a while the Linux kernel was called vmlinux, but the kernel grew too large to fit in the available boot memory, so the kernel image was compressed, and the ending x was changed to a z to show it was compressed with zlib compression. This same compression isn’t always used, often replaced with LZMA or BZIP2, and some kernels are simply called zImage.
At the head of this kernel image (vmlinuz) is a routine that does some minimal amount of hardware setup and then decompresses the kernel contained within the kernel image and places it into high memory. If an initial RAM disk image (initrd) is present, this routine moves it into memory (or we can say extract the compressed ramdisk image into the real memory) and notes it for later use. The routine then calls the kernel, and the kernel boot begins. The initial RAM disk (initrd) is an initial root file system that is mounted prior to when the real rootfile system is available. The initrd is bound to the kernel and loaded as part of the kernel boot procedure. The kernel then mounts this initrd as part of the two-stage boot process to load the modules to make the real file systems available and get at the real root file system. The initrd contains a minimal set of directories and executables to achieve this, such as the insmod tool to install kernel modules into the kernel.
Many Linux distributions ship a single, generic Linux kernel image – one that the distribution's developers create specifically to boot on a wide variety of hardware. The device drivers for this generic kernel image are included as loadable kernel modules because statically compiling many drivers into one kernel causes the kernel image to be much larger, perhaps too large to boot on computers with limited memory.
Create the file /etc/crio/crio.conf.d/50-kata
cat > /etc/crio/crio.conf.d/50-kata << EOF
[crio.runtime.runtimes.kata]
runtime_path = "/usr/local/bin/containerd-shim-kata-v2"
runtime_root = "/run/vc"
runtime_type = "vm"
privileged_without_host_devices = true
EOF
Restart crio and start microshift
systemctl restart crio
systemctl start microshift
export KUBECONFIG=/var/lib/microshift/resources/kubeadmin/kubeconfig
Running some Kata samples
After MicroShift is started, you can apply the kata runtimeclass and run the samples.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/kata/
oc apply -f kata-runtimeclass.yaml
# Start three kata pods
oc apply -f kata-nginx.yaml -f kata-alpine.yaml -f kata-busybox.yaml
watch "oc get nodes;oc get pods -A;crictl stats -a"
Output:
NAME STATUS ROLES AGE VERSION
microshift.example.com Ready <none> 19h v1.21.0
NAMESPACE NAME READY STATUS RESTARTS AGE
default busybox-1 1/1 Running 0 94s
default kata-alpine 1/1 Running 0 94s
default kata-nginx 1/1 Running 0 94s
kube-system console-deployment-768d7dc869-98r5k 1/1 Running 0 17h
kube-system kube-flannel-ds-s4hxb 1/1 Running 0 19h
kubevirt-hostpath-provisioner kubevirt-hostpath-provisioner-l8vtl 1/1 Running 0 19h
openshift-dns dns-default-dx677 2/2 Running 0 19h
openshift-dns node-resolver-5kprs 1/1 Running 0 19h
openshift-ingress router-default-85bcfdd948-72llb 1/1 Running 0 19h
openshift-service-ca service-ca-7764c85869-phtkw 1/1 Running 0 19h
CONTAINER CPU % MEM DISK INODES
003c4997d33d9 0.00 458.8kB 12B 16
0488ddd1ab858 0.00 0B 12B 7
1c91ca2f48c0b 0.00 0B 13.35kB 22
55d22d7d816d0 0.00 0B 6.969kB 11
88705e9cce543 0.00 0B 0B 4
b4cf5fdc56077 1.26 704.5kB 89B 8
d0733e7bd2a99 0.00 0B 12B 18
e353d5e805e36 0.00 0B 0B 3
eca8f7f3ae8a2 0.00 0B 138B 15
f10185631fc64 0.00 2.449MB 1.225kB 21
f6b155e057a37 0.00 0B 12B 19
We can see that the kernel used in the kata containers (when we created the initrd image is 5.4.60) is different from that host 5.15.74 on Host
[root@microshift kata]# kata-runtime kata-env | grep "\[Kernel\]\|\[Host\]" -A 2
[Kernel]
Path = "/usr/share/kata-containers/vmlinux-5.4.60-92"
Parameters = "scsi_mod.scan=none"
--
[Host]
Kernel = "5.15.74-v8.1.el9"
Architecture = "arm64"
[root@microshift kata]# oc exec -it kata-nginx -- uname -a
Linux kata-nginx 5.4.60 #1 SMP Thu Nov 24 14:33:21 UTC 2022 aarch64 GNU/Linux
[root@microshift kata]# oc exec -it kata-alpine -- uname -a
Linux kata-alpine 5.4.60 #1 SMP Thu Nov 24 14:33:21 UTC 2022 aarch64 Linux
[root@microshift kata]# oc exec -it busybox-1 -- uname -a
Linux busybox-1 5.4.60 #1 SMP Thu Nov 24 14:33:21 UTC 2022 aarch64 GNU/Linux
[root@microshift kata]# uname -a
Linux microshift.example.com 5.15.74-v8.1.el9 #1 SMP PREEMPT Thu Oct 27 08:34:02 UTC 2022 aarch64 aarch64 aarch64 GNU/Linux
Try to run ping from the kata-alpine container, it does not work because the capabilities are not set
[root@microshift kata]# oc exec -it kata-alpine -- ping -c2 google.com
PING google.com (142.251.40.206): 56 data bytes
ping: permission denied (are you root?)
command terminated with exit code 1
[root@rpi kata]# oc exec -it busybox-1 -- ping 10-42-0-6.default.pod.cluster.local
PING 10-42-0-6.default.pod.cluster.local (10.42.0.6): 56 data bytes
ping: permission denied (are you root?)
command terminated with exit code 1
Add the registries line under [crio.image] for default registry to download an image from and the NET_RAW to the default_capabilities to /etc/crio/crio.conf so that we can run ping from within the containers. Delete the kata containers, restart crio and microshift and recreate the kata containers.
[crio]
[crio.image]
registries=["quay.io","docker.io"]
# The crio.runtime table contains settings pertaining to the OCI runtime used
# and options for how to set up and manage the OCI runtime.
[crio.runtime]
default_capabilities = [
"CHOWN",
"DAC_OVERRIDE",
"FSETID",
"FOWNER",
"SETGID",
"SETUID",
"SETPCAP",
"NET_BIND_SERVICE",
"KILL",
"NET_RAW",
]
# If true, SELinux will be used for pod separation on the host.
selinux = true
Check that we can ping from the kata-alpine container if you set the default_capabilities "NET_RAW" in /etc/crio/crio.conf as shown in Part 23
[root@microshift kata]# oc delete -f kata-alpine.yaml -f kata-busybox.yaml -f kata-nginx.yaml
pod "kata-alpine" deleted
pod "busybox-1" deleted
pod "kata-nginx" deleted
[root@microshift kata]# systemctl restart crio
[root@microshift kata]# systemctl restart microshift
[root@microshift kata]# watch "oc get nodes;oc get pods -A;crictl stats -a"
[root@microshift kata]# oc apply -f kata-alpine.yaml
pod/kata-alpine created
[root@microshift kata]# oc exec -it kata-alpine -- ping -c2 google.com
PING google.com (142.250.176.206): 56 data bytes
64 bytes from 142.250.176.206: seq=0 ttl=59 time=5.930 ms
64 bytes from 142.250.176.206: seq=1 ttl=59 time=4.972 ms
--- google.com ping statistics ---
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max = 4.972/5.451/5.930 ms
When done, you may delete the sample deployments
[root@microshift kata]# oc delete -f kata-alpine.yaml
pod "kata-alpine" deleted
Influxdb Sample
We execute the runall-balena-dynamic.sh for AlmaLinux 9 after updating the deployment yamls to use the runtimeclass: kata.
cd ~
git clone https://github.com/thinkahead/microshift.git
cd ~/microshift/raspberry-pi/influxdb/
Update the influxdb-deployment.yaml, telegraf-deployment.yaml and grafana/grafana-deployment.yaml to use the runtimeClassName: kata. With Kata containers, we do not directly get access to the host devices. So, we run the measure container as a runc pod. In runc, '--privileged' for a container means all the /dev/* block devices from the host are mounted into the guest. This will allow the privileged container to gain access to mount any block device from the host.
sed -i '/^ spec:/a \ \ \ \ \ \ runtimeClassName: kata' influxdb-deployment.yaml telegraf-deployment.yaml grafana/grafana-deployment.yaml
Now, get the nodename
[root@microshift influxdb]# oc get nodes
NAME STATUS ROLES AGE VERSION
microshift.example.com Ready <none> 19h v1.21.0
Replace the annotation kubevirt.io/provisionOnNode with the above nodename rpi.example.com and execute the runall-balena-dynamic.sh. This will create a new project influxdb.
nodename=microshift.example.com
sed -i "s|kubevirt.io/provisionOnNode:.*| kubevirt.io/provisionOnNode: $nodename|" influxdb-data-dynamic.yaml
sed -i "s| kubevirt.io/provisionOnNode:.*| kubevirt.io/provisionOnNode: $nodename|" grafana/grafana-data-dynamic.yaml
./runall-balena-dynamic.sh
Let’s watch the stats (CPU%, Memory, Disk and Inodes) of the kata container pods:
watch "oc get nodes;oc get pods;crictl stats"
Output:
NAME STATUS ROLES AGE VERSION
microshift.example.com Ready <none> 20h v1.21.0
NAME READY STATUS RESTARTS AGE
grafana-855ffb48d8-z9m4q 1/1 Running 0 16m
influxdb-deployment-6d898b7b7b-tts8x 1/1 Running 0 17m
measure-deployment-58cddb5745-22mm2 1/1 Running 2 17m
telegraf-deployment-d746f5c6-rfvcp 1/1 Running 0 16m
CONTAINER CPU % MEM DISK INODES
160167964a6bb 0.05 24.35MB 4.026MB 70
1f3424266a3bf 0.10 12.08MB 186kB 11
ed155feea1586 0.79 20.58MB 265B 11
We can look at the RUNTIME_CLASS using custom columns:
[root@microshift influxdb]# oc get pods -o custom-columns=NAME:metadata.name,STATUS:.status.phase,RUNTIME_CLASS:.spec.runtimeClassName,IP:.status.podIP,IMAGE:.status.containerStatuses[].image -A
NAME STATUS RUNTIME_CLASS IP IMAGE
grafana-855ffb48d8-z9m4q Running kata 10.42.0.16 docker.io/grafana/grafana:5.4.3
influxdb-deployment-6d898b7b7b-tts8x Running kata 10.42.0.13 docker.io/library/influxdb:1.7.4
measure-deployment-58cddb5745-22mm2 Running <none> 10.42.0.14 docker.io/karve/measure:latest
telegraf-deployment-d746f5c6-rfvcp Running kata 10.42.0.15 docker.io/library/telegraf:1.10.0
console-deployment-768d7dc869-98r5k Running <none> 10.42.0.5 docker.io/karve/console:latest
kube-flannel-ds-s4hxb Running <none> 192.168.1.209 quay.io/microshift/flannel:4.8.0-0.okd-2021-10-10-030117
kubevirt-hostpath-provisioner-l8vtl Running <none> 10.42.0.2 quay.io/microshift/hostpath-provisioner:4.8.0-0.okd-2021-10-10-030117
dns-default-dx677 Running <none> 10.42.0.3 quay.io/microshift/coredns:4.8.0-0.okd-2021-10-10-030117
node-resolver-5kprs Running <none> 192.168.1.209 quay.io/microshift/cli:4.8.0-0.okd-2021-10-10-030117
router-default-85bcfdd948-72llb Running <none> 192.168.1.209 quay.io/microshift/haproxy-router:4.8.0-0.okd-2021-10-10-030117
service-ca-7764c85869-phtkw Running <none> 10.42.0.4 quay.io/microshift/service-ca-operator:4.8.0-0.okd-2021-10-10-030117
Check the qemu process. We used the initrd image and you can see that in the parameters:
ps -ef | grep qemu | grep initrd
Output:
root 222931 1 2 16:18 ? 00:00:30 /usr/libexec/qemu-kvm -name sandbox-1158f4bbe993c6de6e620ba6eb0e1e485441ac329c823990b04ba86865b78147 -uuid 292b9c8e-04cc-4741-917a-09881a010bdd -machine virt,usb=off,accel=kvm,gic-version=host -cpu host,pmu=off -qmp unix:/run/vc/vm/1158f4bbe993c6de6e620ba6eb0e1e485441ac329c823990b04ba86865b78147/qmp.sock,server=on,wait=off -m 2048M,slots=10,maxmem=7812M -device pci-bridge,bus=pcie.0,id=pci-bridge-0,chassis_nr=1,shpc=off,addr=2,io-reserve=4k,mem-reserve=1m,pref64-reserve=1m -device virtio-serial-pci,disable-modern=false,id=serial0 -device virtconsole,chardev=charconsole0,id=console0 -chardev socket,id=charconsole0,path=/run/vc/vm/1158f4bbe993c6de6e620ba6eb0e1e485441ac329c823990b04ba86865b78147/console.sock,server=on,wait=off -device virtio-scsi-pci,id=scsi0,disable-modern=false -object rng-random,id=rng0,filename=/dev/urandom -device virtio-rng-pci,rng=rng0 -device vhost-vsock-pci,disable-modern=false,vhostfd=3,id=vsock-3224903307,guest-cid=3224903307 -chardev socket,id=char-1312d5ca71445b71,path=/run/vc/vm/1158f4bbe993c6de6e620ba6eb0e1e485441ac329c823990b04ba86865b78147/vhost-fs.sock -device vhost-user-fs-pci,chardev=char-1312d5ca71445b71,tag=kataShared,queue-size=1024 -netdev tap,id=network-0,vhost=on,vhostfds=4,fds=5 -device driver=virtio-net-pci,netdev=network-0,mac=26:b4:88:ca:fd:5a,disable-modern=false,mq=on,vectors=4 -rtc base=utc,driftfix=slew,clock=host -global kvm-pit.lost_tick_policy=discard -vga none -no-user-config -nodefaults -nographic --no-reboot -daemonize -object memory-backend-file,id=dimm1,size=2048M,mem-path=/dev/shm,share=on -numa node,memdev=dimm1 -kernel /usr/share/kata-containers/vmlinux-5.4.60-92 -initrd /usr/share/kata-containers/kata-containers-initrd-2022-11-24-12:48:10.440333043+0000-9bde32daa -append iommu.passthrough=0 console=hvc0 console=hvc1 quiet panic=1 nr_cpus=4 scsi_mod.scan=none -pidfile /run/vc/vm/1158f4bbe993c6de6e620ba6eb0e1e485441ac329c823990b04ba86865b78147/pid -smp 1,cores=1,threads=1,sockets=4,maxcpus=4
Add the "<RaspberryPiIPAddress> grafana-service-influxdb.cluster.local" to /etc/hosts on your laptop and login to http://grafana-service-influxdb.cluster.local/login using admin/admin. You may change the password on first login or click skip. Go to the Dashboards list (left menu > Dashboards > Manage). Open the Analysis Server dashboard to display monitoring information for MicroShift. Open the Balena Sense dashboard to show the temperature, pressure, and humidity from SenseHat.
Finally, after you are done working with this sample, you can run the deleteall-balena-dynamic.sh
./deleteall-balena-dynamic.sh
Deleting the persistent volume claims automatically deletes the persistent volumes.
Configure to use the rootfs image
We have been using the initrd image when running the samples above, now let’s switch to the rootfs image instead of using initrd by changing the following lines in /etc/kata-containers/configuration.toml
image = "/usr/share/kata-containers/kata-containers.img"
#initrd = "/usr/share/kata-containers/kata-containers-initrd.img"
Also disable the image nvdimm by setting the following:
disable_image_nvdimm = true # Default is false
Restart crio and test with the kata-alpine sample
systemctl restart crio
cd ~/microshift/raspberry-pi/kata/
oc apply -f kata-alpine.yaml
Output of qemu process when we use rootfs image with disable_image_nvdimm=true
root 233569 1 28 16:44 ? 00:00:05 /usr/libexec/qemu-kvm -name sandbox-e8806511f0f6564d56417fb96a17f354946ce0f5fc783aeb38e09030124d0d7b -uuid b07b2f1d-0f7a-4d96-b712-2142ca0bb0bf -machine virt,usb=off,accel=kvm,gic-version=host -cpu host,pmu=off -qmp unix:/run/vc/vm/e8806511f0f6564d56417fb96a17f354946ce0f5fc783aeb38e09030124d0d7b/qmp.sock,server=on,wait=off -m 2048M,slots=10,maxmem=7812M -device pci-bridge,bus=pcie.0,id=pci-bridge-0,chassis_nr=1,shpc=off,addr=2,io-reserve=4k,mem-reserve=1m,pref64-reserve=1m -device virtio-serial-pci,disable-modern=false,id=serial0 -device virtconsole,chardev=charconsole0,id=console0 -chardev socket,id=charconsole0,path=/run/vc/vm/e8806511f0f6564d56417fb96a17f354946ce0f5fc783aeb38e09030124d0d7b/console.sock,server=on,wait=off -device virtio-blk-pci,disable-modern=false,drive=image-91f5e6bf52b9981d,scsi=off,config-wce=off,share-rw=on,serial=image-91f5e6bf52b9981d -drive id=image-91f5e6bf52b9981d,file=/usr/share/kata-containers/kata-containers-2022-11-24-14:19:58.932179446+0000-9bde32daa,aio=threads,format=raw,if=none,readonly=on -device virtio-scsi-pci,id=scsi0,disable-modern=false -object rng-random,id=rng0,filename=/dev/urandom -device virtio-rng-pci,rng=rng0 -device vhost-vsock-pci,disable-modern=false,vhostfd=3,id=vsock-699490259,guest-cid=699490259 -chardev socket,id=char-d491e2de85516dea,path=/run/vc/vm/e8806511f0f6564d56417fb96a17f354946ce0f5fc783aeb38e09030124d0d7b/vhost-fs.sock -device vhost-user-fs-pci,chardev=char-d491e2de85516dea,tag=kataShared,queue-size=1024 -netdev tap,id=network-0,vhost=on,vhostfds=4,fds=5 -device driver=virtio-net-pci,netdev=network-0,mac=2e:ed:25:a4:da:c2,disable-modern=false,mq=on,vectors=4 -rtc base=utc,driftfix=slew,clock=host -global kvm-pit.lost_tick_policy=discard -vga none -no-user-config -nodefaults -nographic --no-reboot -daemonize -object memory-backend-file,id=dimm1,size=2048M,mem-path=/dev/shm,share=on -numa node,memdev=dimm1 -kernel /usr/share/kata-containers/vmlinux-5.4.60-92 -append iommu.passthrough=0 root=/dev/vda1 rootflags=data=ordered,errors=remount-ro ro rootfstype=ext4 console=hvc0 console=hvc1 quiet systemd.show_status=false panic=1 nr_cpus=4 systemd.unit=kata-containers.target systemd.mask=systemd-networkd.service systemd.mask=systemd-networkd.socket scsi_mod.scan=none -pidfile /run/vc/vm/e8806511f0f6564d56417fb96a17f354946ce0f5fc783aeb38e09030124d0d7b/pid -smp 1,cores=1,threads=1,sockets=4,maxcpus=4
We can also run MicroShift Containerized as shown in Part 18 and execute the Jupyter Notebook samples for Digit Recognition, Object Detection and License Plate Recognition with Kata containers as shown in Part 23.
Error Messages in MicroShift Logs
Message: failed to fetch hugetlb info
The following journalctl command will continuously show warning messages “failed to fetch hugetlb info”. The default kernel for the Raspberry Pi OS does not support HugeTLB hugepages.
journalctl -u microshift -f
Output:
...
May 23 11:49:20 rpi.example.com microshift[3112]: W0523 11:49:20.347604 3112 container.go:586] Failed to update stats for container "/system.slice/crio-00c0b63eeee509979e7652cca8b91a1e9e900c1989b7fe54f6a05f6591de0108.scope": error while statting cgroup v2: [open /sys/kernel/mm/hugepages: no such file or directory
May 23 11:49:20 rpi.example.com microshift[3112]: failed to fetch hugetlb info
...
To remove these messages, we can recompile the microshift binary using the changes from hugetlb.go.
dnf -y install wget pkg-config
# Install golang 1.17.2 (Do not use 1.18.x or 1.19.x)
wget https://golang.org/dl/go1.17.2.linux-arm64.tar.gz
rm -rf /usr/local/go && tar -C /usr/local -xzf go1.17.2.linux-arm64.tar.gz
rm -f go1.17.2.linux-arm64.tar.gz
export PATH=$PATH:/usr/local/go/bin
export GOPATH=/root/go
cat << EOF >> /root/.bashrc
export PATH=$PATH:/usr/local/go/bin
export GOPATH=/root/go
EOF
mkdir $GOPATH
git clone https://github.com/thinkahead/microshift.git
cd microshift
Edit the file vendor/github.com/opencontainers/runc/libcontainer/cgroups/fs2/hugetlb.go and remove the return on err.
func statHugeTlb(dirPath string, stats *cgroups.Stats) error {
hugePageSizes, _ := cgroups.GetHugePageSize()
//hugePageSizes, err := cgroups.GetHugePageSize()
//if err != nil {
// return errors.Wrap(err, "failed to fetch hugetlb info")
//}
hugetlbStats := cgroups.HugetlbStats{}
Build and replace the microshift binary. Restart MicroShift.
make
# Check which binary the /usr/lib/systemd/system/microshift.service points to
mv microshift /usr/bin/.
#mv microshift /usr/local/bin/.
restorecon -R -v /usr/bin/microshift # See below
systemctl restart microshift
Permission denied in journalctl microshift logs when you restart microshift
You will see the following errors when you copy the microshift binary over to /usr/bin
[root@microshift microshift]# systemctl restart microshift
[root@microshift microshift]# journalctl -u microshift -f
Nov 23 21:05:20 microshift.example.com systemd[1]: Started MicroShift.
Nov 23 21:05:20 microshift.example.com systemd[53453]: microshift.service: Failed to locate executable /usr/bin/microshift: Permission denied
Nov 23 21:05:20 microshift.example.com systemd[53453]: microshift.service: Failed at step EXEC spawning /usr/bin/microshift: Permission denied
Fix it by restoring the default context of /usr/bin/microshift
[root@microshift microshift]# ls -Z /usr/bin/microshift
unconfined_u:object_r:user_tmp_t:s0 /usr/bin/microshift
[root@microshift microshift]# restorecon -R -v /usr/bin/microshift
Relabeled /usr/bin/microshift from unconfined_u:object_r:user_tmp_t:s0 to unconfined_u:object_r:container_runtime_exec_t:s0
[root@microshift microshift]# ls -Z /usr/bin/microshift
unconfined_u:object_r:container_runtime_exec_t:s0 /usr/bin/microshift
Unable to fetch pod log stats
Deleted pods are not properly reaped on nodes. The observed behavior is that after deletion, we see the pods added to a list of recurring events that keeps spamming the logs. This happens with CRI-O >= 1.22.
Nov 23 22:55:51 microshift.example.com microshift[54264]: E1123 22:55:51.898173 54264 cadvisor_stats_provider.go:147] "Unable to fetch pod log stats" err="open /var/log/pods/kubevirt_kubevirt-3993aff5e8e705c86a9c9145f9953a43e32e0468-jobzz9pgwd7g5_9c9d77aa-5feb-45e1-9364-6aac7357b337: no such file or directory" pod="kubevirt/kubevirt-3993aff5e8e705c86a9c9145f9953a43e32e0468-jobzz9pgwd7g5"
Nov 23 22:55:51 microshift.example.com microshift[54264]: E1123 22:55:51.898695 54264 cadvisor_stats_provider.go:147] "Unable to fetch pod log stats" err="open /var/log/pods/influxdb_measure-deployment-58cddb5745-7nkmv_46948320-778e-4f3c-8246-20dba7c6bd44: no such file or directory" pod="influxdb/measure-deployment-58cddb5745-7nkmv"
Nov 23 22:55:51 microshift.example.com microshift[54264]: E1123 22:55:51.899081 54264 cadvisor_stats_provider.go:147] "Unable to fetch pod log stats" err="open /var/log/pods/kubevirt_virt-operator-849f75c557-c5njq_4e482dc8-6997-44d5-880e-c2bf0bf499ab: no such file or directory" pod="kubevirt/virt-operator-849f75c557-c5njq"
The fix will require making changes as mentioned at https://github.com/kubernetes/kubernetes/pull/108855/files and rebuilding the microshift binary.
Conclusion
In this Part 26, we saw multiple options to run MicroShift on the Raspberry Pi 4 with the AlmaLinux 9 (64 bit). We used dynamic persistent volumes to install InfluxDB/Telegraf/Grafana with a dashboard to show SenseHat sensor data. We ran samples that used the Sense Hat/USB camera and worked with a sample that sent the pictures and web socket messages to Node Red when a person was detected. We ran a sample with MongoDB. We created Virtual Machines, installed and used CDI to upload images, installed the OKD Web Console and saw how to connect to a Virtual Machine Instance using KubeVirt. Finally, we built and ran Kata containers. In the next Part 27 we will work with Oracle Linux 9.
Hope you have enjoyed the article. Share your thoughts in the comments or engage in the conversation with me on Twitter @aakarve. I look forward to hearing about your use of MicroShift on ARM devices and if you would like to see something covered in more detail.
References
#MicroShift#Openshift#containers#crio#Edge#raspberry-pi#almalinux #mongodb #cdi