# List snapshots used by the pod's namespace nerdctl -n k8s.io ps -a # Get snapshot size directly (if using buildkit) nerdctl -n k8s.io image ls -a If you find orphaned overlay mounts ( findmnt | grep overlay shows many old pods):
Run: ps aux | grep -E "containerd|crio|dockerd" We will cover installation on Ubuntu 22.04/24.04 LTS , RHEL 9 / CentOS 9 , and macOS (for remote debugging) . Part 1: Installing cri-tools (crictl) crictl is the Swiss Army knife. It does not care about your underlying filesystem; it talks to the CRI socket. On Ubuntu/Debian # Add Kubernetes repository (contains cri-tools) sudo apt-get update sudo apt-get install -y apt-transport-https ca-certificates curl curl -fsSL https://pkgs.k8s.io/core:/stable:/v1.30/deb/Release.key | sudo gpg --dearmor -o /etc/apt/keyrings/kubernetes-apt-keyring.gpg echo 'deb [signed-by=/etc/apt/keyrings/kubernetes-apt-keyring.gpg] https://pkgs.k8s.io/core:/stable:/v1.30/deb/ /' | sudo tee /etc/apt/sources.list.d/kubernetes.list sudo apt-get update sudo apt-get install -y cri-tools On RHEL/CentOS/Fedora cat <<EOF | sudo tee /etc/yum.repos.d/kubernetes.repo [kubernetes] name=Kubernetes baseurl=https://pkgs.k8s.io/core:/stable:/v1.30/rpm/ enabled=1 gpgcheck=1 gpgkey=https://pkgs.k8s.io/core:/stable:/v1.30/rpm/repodata/repomd.xml.key EOF sudo yum install -y cri-tools Manual Install (All Linux distros) For the latest version (bypassing package managers):
| Runtime | CRI Socket | Default CLI Tools | | :--- | :--- | :--- | | | /run/containerd/containerd.sock | ctr , nerdctl , crictl | | CRI-O | /run/crio/crio.sock | crictl , podman | | Docker (via cri-dockerd) | /run/cri-dockerd.sock | crictl (limited) | cri file system tools install
nerdctl images nerdctl inspect <image> nerdctl run --rm -it alpine ls / Part 3: Installing Snapshotter Tools (OverlayFS Utilities) To truly debug the CRI filesystem, you need host-level tools that understand overlayfs (the default snapshotter for 99% of clusters).
sudo ls -la /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/23/fs Use du -sh to find the bloat: # List snapshots used by the pod's namespace nerdctl -n k8s
# Download nerdctl full bundle (includes containerd + runc + CNI) curl -LO https://github.com/containerd/nerdctl/releases/download/v1.7.6/nerdctl-full-1.7.6-linux-amd64.tar.gz sudo tar Cxzvvf /usr/local nerdctl-full-1.7.6-linux-amd64.tar.gz curl -LO https://github.com/containerd/nerdctl/releases/download/v1.7.6/nerdctl-1.7.6-linux-amd64.tar.gz sudo tar Cxzvvf /usr/local/bin nerdctl-1.7.6-linux-amd64.tar.gz
In the rapidly evolving landscape of cloud-native development and container orchestration, the underlying infrastructure often becomes an opaque black box. For developers and system administrators working with Kubernetes, containerd, or CRI-O, one acronym repeatedly surfaces when debugging storage issues: CRI . sudo du -sh /var/lib/containerd/io
sudo du -sh /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/*/fs | sort -h The largest directory is the culprit. With nerdctl , you can bypass low-level snapshot IDs: