kata-containers/docs/how-to/how-to-pull-images-in-guest-with-kata.md

Kata Containers with Guest Image Management

Kata Containers 3.3.0 introduces the guest image management feature, which enables the guest VM to directly pull images using nydus snapshotter. This feature is designed to protect the integrity of container images and guard against any tampering by the host, which is used for confidential containers. Please refer to kata-guest-image-management-design for details.

Prerequisites

• The k8s cluster with Kata 3.3.0+ is ready to use.
• yq is installed in the host and it's directory is included in the PATH environment variable. (optional, for DaemonSet only)

Deploy nydus snapshotter for guest image management

To pull images in the guest, we need to do the following steps:

1. Delete images used for pulling in the guest (optional, for containerd only)
2. Install nydus snapshotter:
   1. Install nydus snapshotter by k8s DaemonSet (recommended)
   2. Install nydus snapshotter manually

Delete images used for pulling in the guest

Though the CRI Runtime Specific Snapshotter is still an experimental feature in containerd, which containerd is not supported to manage the same image in different snapshotters(The default snapshotter in containerd is overlayfs). To avoid errors caused by this, it is recommended to delete images (including the pause image) in containerd that needs to be pulled in guest later before configuring nydus snapshotter in containerd.

Install nydus snapshotter

Install nydus snapshotter by k8s DaemonSet (recommended)

To use DaemonSet to install nydus snapshotter, we need to ensure that yq exists in the host.

1. Download nydus snapshotter repo

$ nydus_snapshotter_install_dir="/tmp/nydus-snapshotter"
$ nydus_snapshotter_url=https://github.com/containerd/nydus-snapshotter
$ nydus_snapshotter_version="v0.13.11"
$ git clone -b "${nydus_snapshotter_version}" "${nydus_snapshotter_url}" "${nydus_snapshotter_install_dir}"

2. Configure DaemonSet file

$ pushd "$nydus_snapshotter_install_dir"
$ yq -i \
>	 '.data.FS_DRIVER = "proxy"' -P \
>	 misc/snapshotter/base/nydus-snapshotter.yaml
# Disable to read snapshotter config from configmap
$ yq -i \
>	 'data.ENABLE_CONFIG_FROM_VOLUME = "false"' -P \
>	 misc/snapshotter/base/nydus-snapshotter.yaml
# Enable to run snapshotter as a systemd service 
# (skip if you want to run nydus snapshotter as a standalone process)
$ yq -i \
>	 'data.ENABLE_SYSTEMD_SERVICE = "true"' -P \
>	 misc/snapshotter/base/nydus-snapshotter.yaml
# Enable "runtime specific snapshotter" feature in containerd when configuring containerd for snapshotter
# (skip if you want to configure nydus snapshotter as a global snapshotter in containerd)
$ yq -i \
>	 'data.ENABLE_RUNTIME_SPECIFIC_SNAPSHOTTER = "true"' -P \
>	 misc/snapshotter/base/nydus-snapshotter.yaml

3. Install nydus snapshotter as a DaemonSet

$ kubectl create -f "misc/snapshotter/nydus-snapshotter-rbac.yaml"
$ kubectl apply -f "misc/snapshotter/base/nydus-snapshotter.yaml"

4. Wait 5 minutes until the DaemonSet is running

$ kubectl rollout status DaemonSet nydus-snapshotter -n nydus-system --timeout 5m

5. Verify whether nydus snapshotter is running as a DaemonSet

$ pods_name=$(kubectl get pods --selector=app=nydus-snapshotter -n nydus-system -o=jsonpath='{.items[*].metadata.name}')
$ kubectl logs "${pods_name}" -n nydus-system
deploying snapshotter
install nydus snapshotter artifacts
configuring snapshotter
Not found nydus proxy plugin!
running snapshotter as systemd service
Created symlink /etc/systemd/system/multi-user.target.wants/nydus-snapshotter.service → /etc/systemd/system/nydus-snapshotter.service.

Install nydus snapshotter manually

1. Download nydus snapshotter binary from release

$ ARCH=$(uname -m)
$ golang_arch=$(case "$ARCH" in
    aarch64) echo "arm64" ;;
    ppc64le) echo "ppc64le" ;;
    x86_64) echo "amd64" ;;
    s390x) echo "s390x" ;;
esac)
$ release_tarball="nydus-snapshotter-${nydus_snapshotter_version}-linux-${golang_arch}.tar.gz"
$ curl -OL ${nydus_snapshotter_url}/releases/download/${nydus_snapshotter_version}/${release_tarball}
$ sudo tar -xfz ${release_tarball} -C /usr/local/bin --strip-components=1

2. Download nydus snapshotter configuration file for pulling images in the guest

$ curl -OL https://github.com/containerd/nydus-snapshotter/blob/main/misc/snapshotter/config-proxy.toml
$ sudo install -D -m 644 config-proxy.toml /etc/nydus/config-proxy.toml

3. Run nydus snapshotter as a standalone process

$ /usr/local/bin/containerd-nydus-grpc --config /etc/nydus/config-proxy.toml --log-to-stdout
level=info msg="Start nydus-snapshotter. Version: v0.13.11-308-g106a6cb, PID: 1100169, FsDriver: proxy, DaemonMode: none"
level=info msg="Run daemons monitor..."

4. Configure containerd for nydus snapshotter

Configure nydus snapshotter to enable CRI Runtime Specific Snapshotter in containerd. This ensures run kata containers with nydus snapshotter. Below, the steps are illustrated using kata-qemu as an example.

# Modify containerd configuration to ensure that the following lines appear in the containerd configuration 
# (Assume that the containerd config is located in /etc/containerd/config.toml)

[plugins."io.containerd.grpc.v1.cri".containerd]
  disable_snapshot_annotations = false
  discard_unpacked_layers = false
[proxy_plugins.nydus]
  type = "snapshot"
  address = "/run/containerd-nydus/containerd-nydus-grpc.sock"
[plugins."io.containerd.grpc.v1.cri".containerd.runtimes.kata-qemu]
  snapshotter = "nydus"

Notes: The CRI Runtime Specific Snapshotter feature only works for containerd v1.7.0 and above. So for Containerd v1.7.0 below, in addition to the above settings, we need to set the global snapshotter to nydus in containerd config. For example:

[plugins."io.containerd.grpc.v1.cri".containerd]
snapshotter = "nydus"

5. Restart containerd service

$ sudo systemctl restart containerd

Run pod in kata containers with pulling image in guest

To verify pulling images in a guest VM, please refer to the following commands:

1. Run a kata container

$ cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: Pod
metadata:
  name: busybox
spec:
  runtimeClassName: kata-qemu
  containers:
  - name: busybox
    image: quay.io/prometheus/busybox:latest
    imagePullPolicy: Always
EOF
pod/busybox created
$ kubectl get pods
NAME         READY   STATUS    RESTARTS   AGE
busybox      1/1     Running   0          10s

2. Verify that the pod's images have been successfully downloaded in the guest. If images intended for deployment are deleted prior to deploying with nydus snapshotter, the root filesystems required for the pod's images (including the pause image and the container image) should not be present on the host.

$ sandbox_id=$(ps -ef| grep containerd-shim-kata-v2| grep -oP '(?<=-id\s)[a-f0-9]+'| tail -1)
$ rootfs_count=$(find /run/kata-containers/shared/sandboxes/$sandbox_id -name rootfs -type d| grep -o "rootfs" | wc -l)
$ echo $rootfs_count
0

Run pod in kata containers with pulling large image in guest

Currently, the image pulled in the guest will be downloaded and unpacked in the /run/kata-containers/image directory. However, by default, in rootfs-confidential image, systemd allocates 50% of the available physical RAM to the /run directory using a tmpfs filesystem. As we all know, memory is valuable, especially for confidential containers. This means that if we run a kata container with the default configuration (where the default memory assigned for a VM is 2048 MiB), /run would be allocated around 1024 MiB. Consequently, we can only pull images up to 1024 MiB in the guest. So we can use a block volume from the host and use dm-crypt and dm-integrity to encrypt the block volume in the guest, providing a secure place to store downloaded container images.

Create block volume with k8s

There are a lot of CSI Plugins that support block volumes: AWS EBS, Azure Disk, Open-Local and so on. But as an example, we use Local Persistent Volumes to use local disks as block storage with k8s cluster.

1. Create an empty disk image and attach the image to a loop device, such as /dev/loop0

$ loop_file="/tmp/trusted-image-storage.img"
$ sudo dd if=/dev/zero of=$loop_file bs=1M count=2500
$ sudo losetup /dev/loop0 $loop_file

2. Create a Storage Class

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: local-storage
provisioner: kubernetes.io/no-provisioner
volumeBindingMode: WaitForFirstConsumer

3. Create Persistent Volume

apiVersion: v1
kind: PersistentVolume
metadata:
  name: trusted-block-pv
spec:
  capacity:
    storage: 10Gi
  volumeMode: Block
  accessModes:
    - ReadWriteOnce
  persistentVolumeReclaimPolicy: Retain
  storageClassName: local-storage
  local:
    path: /dev/loop0
  nodeAffinity:
    required:
      nodeSelectorTerms:
      - matchExpressions:
        - key: kubernetes.io/hostname
          operator: In
          values:
          - NODE_NAME

4. Create Persistent Volume Claim

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: trusted-pvc
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 1Gi
  volumeMode: Block
  storageClassName: local-storage

5. Run a pod with pulling large image in guest

apiVersion: v1
kind: Pod
metadata:
  name: large-image-pod
spec:
  runtimeClassName: kata-qemu
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: kubernetes.io/hostname
            operator: In
            values:
            - NODE_NAME
  volumes:
    - name: trusted-storage
      persistentVolumeClaim:
        claimName: trusted-pvc
  containers:
    - name: app-container
      image: quay.io/confidential-containers/test-images:largeimage
      command: ["/bin/sh", "-c"]
      args:
        - sleep 6000
      volumeDevices:
        - devicePath: /dev/trusted_store
          name: trusted-image-storage

5. Docker image size

docker image ls|grep "largeimage"
quay.io/confidential-containers/test-images               largeimage            00bc1f6c893a   4 months ago   2.15GB

6. Check whether the device is encrypted and used by entering into the VM

$ lsblk --fs
NAME                 FSTYPE LABEL UUID FSAVAIL FSUSE% MOUNTPOINT
sda                                                   
└─encrypted_disk_GsLDt
                                          178M    87% /run/kata-containers/image

$ cryptsetup status encrypted_disk_GsLDt
/dev/mapper/encrypted_disk_GsLDt is active and is in use.
  type:    LUKS2
  cipher:  aes-xts-plain64
  keysize: 512 bits
  key location: keyring
  device:  /dev/sda
  sector size:  4096
  offset:  32768 sectors
  size:    5087232 sectors
  mode:    read/write

$ mount|grep "encrypted_disk_GsLDt"
/dev/mapper/encrypted_disk_GsLDt on /run/kata-containers/image type ext4

$ du -h --max-depth=1 /run/kata-containers/image/
16K     /run/kata-containers/image/lost+found
2.1G    /run/kata-containers/image/layers
60K     /run/kata-containers/image/overlay
2.1G    /run/kata-containers/image/

$ free -m
              total        used        free      shared  buff/cache   available
Mem:           1989          52          43           0        1893        1904
Swap:             0           0           0