How to work with container images using ctr
Basic image management with ctr
Pulling images with ctr always requires a fully-qualified reference - in other words, you cannot omit the domain or the tag parts (the digest doesn’t have to be provided, though).
ctr image pull docker.io/library/nginx:latest
ctr image pull quay.io/quay/busybox:latest
Listing local images
ctr image ls
Build and Load images
However, despite the fact the containerd is often used by higher-level tools to build container images, it doesn’t provide out-of-the-box image building functionality, so there’s no ctr image build command.
Luckily, you can load existing images into containerd using ctr image import. To some extent, it can mitigate the absence of the build command. Here is how you can build an image using a traditional docker build command and then import it:
docker build -t example.com/iximiuz/test:latest - <<EOF
FROM busybox:latest
CMD echo just a test
EOF
docker save -o iximiuz-test.tar example.com/iximiuz/test:latest
ctr image import iximiuz-test.tar
Tag images
ctr image tag example.com/iximiuz/test:latest localhost:5000/iximiuz/test:latest
Push images
ctr image push localhost:5000/iximiuz/test:latest
Remove images
ctr image remove localhost:5000/iximiuz/test:latest
Export and Inspect images
Sometimes, you may want to inspect the internals of an image. With ctr, you can do it using the ctr image export command that saves the image’s content to a tarball:
ctr image export /tmp/nginx.tar docker.io/library/nginx:latest
The exported tarball will contain the nginx image data stored using the OCI Image Layout format.
You can extract the tarball to a temporary directory and explore its contents:
mkdir /tmp/nginx_image
tar -xf /tmp/nginx.tar -C /tmp/nginx_image/
ls -lah /tmp/nginx_image/
The result should look as follows:
total 24K
drwxr-xr-x 3 root root 4.0K Jun 3 11:23 .
drwxrwxrwt 11 root root 4.0K Jun 3 11:23 ..
drwxr-xr-x 3 root root 4.0K Jan 1 1970 blobs
-rw-r--r-- 1 root root 323 Jan 1 1970 index.json
-rw-r--r-- 1 root root 611 Jan 1 1970 manifest.json
-r--r--r-- 1 root root 30 Jan 1 1970 oci-layout
When dealing with raw image layers is not desirable, you can mount the image to a temporary directory on the host and explore its root filesystem as it were a regular directory:
mkdir /tmp/nginx_rootfs
ctr image mount docker.io/library/nginx:latest /tmp/nginx_rootfs
The file and directories of the Nginx image should be available under /tmp/nginx_rootfs path now. Try:
ls -l /tmp/nginx_rootfs/
The result will look like a typical Linux filesystem, and it’s not surprising since the Nginx image is based on the Debian Linux distribution:
total 84
drwxr-xr-x 2 root root 4096 May 2 00:00 bin
drwxr-xr-x 2 root root 4096 Apr 2 11:55 boot
drwxr-xr-x 2 root root 4096 May 2 00:00 dev
drwxr-xr-x 1 root root 4096 May 3 19:51 docker-entrypoint.d
-rwxrwxr-x 1 root root 1616 May 3 19:50 docker-entrypoint.sh
drwxr-xr-x 1 root root 4096 May 3 19:51 etc
...
Don’t forget to unmount the image when you’re done with it
Source:
https://labs.iximiuz.com/courses/containerd-cli/ctr/image-management