README
¶
img
Standalone, daemon-less, unprivileged Dockerfile and OCI compatible container image builder.
img is more cache-efficient than Docker and can also execute multiple build stages concurrently,
as it internally uses BuildKit's DAG solver.
The commands/UX are the same as docker {build,push,pull,login} so all you
have to do is replace docker with img in your scripts, command line, and/or life.
You might also be interested in reading:
Currently you can run it as unprivileged and there are a few ways to go about
this, the default backend naive will not mount if you are running as an
unprivileged user and so it should work. You can learn more about the
snapshotter backends here.
However, it does not currently work as an unprivileged user in a container
due to the fact that it cant mount proc inside the container. See
opencontainers/runc#1658
for more info.
But it will work as an unprivileged user on your host with a runc that is
compiled from the following branch: AkihiroSuda/runc/tree/demo-rootless. I also uploaded one here: https://misc.j3ss.co/tmp/runc, if you trust me.
TLDR; will work unprivileged on your host not in a container til we work out some kinks.
Acknowledgements
A lot of this is based on the work of moby/buildkit. Thanks @tonistiigi and @AkihiroSuda!
Prior Art
The best one you should be using is cyphar/orca-build.
You can already do the same thing as img today with skopeo and umoci. This is just a hack on top of buildkit. I thought it was fun to try a FUSE snapshotter and then I saw @AkihiroSuda's runc patches for making buildkit rootless and thought it would be fun to use as well.
THIS IS NOT NOVEL.
You could even probably use buildah as unprivileged if you use the same instructions from the unprivileged mounting section below.
OR you can read this blog post and use skopeo, ostree and runc.
Table of Contents
Installation
You need to have runc, execv, and newuidmap installed. On ubuntu the former two are provided by the packages
setpriv and uidmap
For the FUSE backend, you will also need fusermount installed.
NOTE: These steps work only for Linux. Compile and run in a container (explained below) if you're on Windows or MacOS.
Binaries
- linux amd64
Via Go
$ go get github.com/genuinetools/img
From Source
$ mkdir -p $GOPATH/src/github.com/genuinetools
$ git clone https://github.com/genuinetools/img $GOPATH/src/github.com/genuinetools/img
$ cd !$
$ make
$ sudo make install
Running with Docker
$ docker run --rm -it \
--name img \
--volume $(pwd):/home/user/src:ro \ # for the build context and dockerfile, can be read-only since we won't modify it
--workdir /home/user/src \ # set the builder working directory
--volume "${HOME}/.docker:/root/.docker:ro" \ # for credentials to push to docker hub or a registry
--cap-add SETGID \ # so we can set groups
--cap-add SETUID \ # so we can map users in a user namespace
--security-opt apparmor=unconfined \ # turn off apparmor so we can mount unprivileged
--security-opt seccomp=unconfined \ # turn off seccomp because it blocks new user namespaces
r.j3ss.co/img build -t user/myimage .
Usage
$ img -h
Usage: img <command>
Commands:
build Build an image from a Dockerfile.
du Show image disk usage.
ls List images and digests.
login Log in to a Docker registry.
pull Pull an image or a repository from a registry.
push Push an image or a repository to a registry.
rm Remove one or more images.
save Save an image to a tar archive (streamed to STDOUT by default).
tag Create a tag TARGET_IMAGE that refers to SOURCE_IMAGE.
version Show the version information.
Build an Image
$ img build -h
Usage: img build [OPTIONS] PATH
Build an image from a Dockerfile.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-build-arg Set build-time variables (default: [])
-d enable debug logging (default: false)
-f Name of the Dockerfile (Default is 'PATH/Dockerfile') (default: <none>)
-state directory to hold the global state (default: /tmp/img)
-t Name and optionally a tag in the 'name:tag' format (default: <none>)
-target Set the target build stage to build (default: <none>)
Use just like you would docker build.
$ img build -t jess/img .
Building jess/img
Setting up the rootfs... this may take a bit.
RUN [/bin/sh -c apk add --no-cache ca-certificates]
--->
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/community/x86_64/APKINDEX.tar.gz
OK: 5 MiB in 12 packages
<--- 5e433zdbh8eosea0u9b70axb3 0 <nil>
RUN [copy /src-0 /dest/go/src/github.com/genuinetools/img]
--->
<--- rqku3imaivvjpgl676se1gupc 0 <nil>
RUN [/bin/sh -c set -x && apk add --no-cache --virtual .build-deps bash git gcc libc-dev libgcc libseccomp-dev linux-headers make && cd /go/src/github.com/genuinetools/img && make static && mv img /usr/bin/img && mkdir -p /go/src/github.com/opencontainers && git clone https://github.com/opencontainers/runc /go/src/github.com/opencontainers/runc && cd /go/src/github.com/opencontainers/runc && make static BUILDTAGS="seccomp" EXTRA_FLAGS="-buildmode pie" EXTRA_LDFLAGS="-extldflags \\\"-fno-PIC -static\\\"" && mv runc /usr/bin/runc && apk del .build-deps && rm -rf /go && echo "Build complete."]
--->
+ apk add --no-cache --virtual .build-deps bash git gcc libc-dev libgcc libseccomp-dev linux-headers make
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/community/x86_64/APKINDEX.tar.gz
(1/28) Installing pkgconf (1.3.10-r0)
(2/28) Installing ncurses-terminfo-base (6.0_p20171125-r0)
(3/28) Installing ncurses-terminfo (6.0_p20171125-r0)
(4/28) Installing ncurses-libs (6.0_p20171125-r0)
(5/28) Installing readline (7.0.003-r0)
(6/28) Installing bash (4.4.19-r1)
....
....
RUN [copy /src-0/certs /dest/etc/ssl/certs]
--->
<--- 6ljir2x800w6deqlradhw0dy2 0 <nil>
Successfully built jess/img
List Image Layers
$ img ls -h
Usage: img ls [OPTIONS]
List images and digests.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-f Filter output based on conditions provided (default: [])
-state directory to hold the global state (default: /tmp/img)
$ img ls
NAME SIZE CREATED AT UPDATED AT DIGEST
jess/img:latest 1.534KiB 9 seconds ago 9 seconds ago sha256:27d862ac32022946d61afbb91ddfc6a1fa2341a78a0da11ff9595a85f651d51e
jess/thing:latest 591B 30 minutes ago 30 minutes ago sha256:d664b4e9b9cd8b3067e122ef68180e95dd4494fd4cb01d05632b6e77ce19118e
Pull an Image
$ img pull -h
Usage: img pull [OPTIONS] NAME[:TAG|@DIGEST]
Pull an image or a repository from a registry.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-state directory to hold the global state (default: /tmp/img)
$ img pull r.j3ss.co/stress
Pulling r.j3ss.co/stress:latest...
Snapshot ref: sha256:2bb7a0a5f074ffe898b1ef64b3761e7f5062c3bdfe9947960e6db48a998ae1d6
Size: 365.9KiB
Push an Image
$ img push -h
Usage: img push [OPTIONS] NAME[:TAG]
Push an image or a repository to a registry.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-state directory to hold the global state (default: /tmp/img)
$ img push jess/thing
Pushing jess/thing:latest...
Successfully pushed jess/thing:latest
Tag an Image
$ img tag -h
Usage: img tag SOURCE_IMAGE[:TAG] TARGET_IMAGE[:TAG]
Create a tag TARGET_IMAGE that refers to SOURCE_IMAGE.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-state directory to hold the global state (default: /tmp/img)
$ img tag jess/thing jess/otherthing
Successfully tagged jess/thing as jess/otherthing
Export an Image to Docker
$ img save -h
Usage: img save [OPTIONS] IMAGE [IMAGE...]
Save an image to a tar archive (streamed to STDOUT by default).
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-o Write to a file, instead of STDOUT (default: <none>)
-state directory to hold the global state (default: /tmp/img)
$ img save jess/thing | docker load
6c3d70c8619c: Loading layer [==================================================>] 9.927MB/9.927MB
7e336c441b5e: Loading layer [==================================================>] 5.287MB/5.287MB
533fecff21a8: Loading layer [==================================================>] 2.56MB/2.56MB
3db7019eac28: Loading layer [==================================================>] 1.679kB/1.679kB
Loaded image: jess/thing
Remove an Image
Usage: img rm [OPTIONS] IMAGE [IMAGE...]
Remove one or more images.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-state directory to hold the global state (default: /tmp/img)
Disk Usage
$ img du -h
Usage: img du [OPTIONS]
Show image disk usage.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-f Filter output based on conditions provided (snapshot ID supported) (default: <none>)
-state directory to hold the global state (default: /tmp/img)
$ img du
ID RECLAIMABLE SIZE DESCRIPTION
sha256:d9a48086f223d28a838263a6c04705c8009fab1dd67cc82c0ee821545de3bf7c true 911.8KiB pulled from docker.io/tonistiigi/copy@sha256:476e0a67a1e4650c6adaf213269a2913deb7c52cbc77f954026f769d51e1a14e
7ia86xm2e4hzn2u947iqh9ph2 true 203.2MiB mount /dest from exec copy /src-0 /dest/go/src/github.com/genuinetools/img
...
sha256:9f131fba0383a6aaf25ecd78bd5f37003e41a4385d7f38c3b0cde352ad7676da true 958.6KiB pulled from docker.io/library/golang:alpine@sha256:a0045fbb52a7ef318937e84cf7ad3301b4d2ba6cecc2d01804f428a1e39d1dfc
sha256:c4151b5a5de5b7e272b2b6a3a4518c980d6e7f580f39c85370330a1bff5821f1 true 472.3KiB pulled from docker.io/tonistiigi/copy@sha256:476e0a67a1e4650c6adaf213269a2913deb7c52cbc77f954026f769d51e1a14e
sha256:ae4ecac23119cc920f9e44847334815d32bdf82f6678069d8a8be103c1ee2891 true 148.9MiB pulled from docker.io/library/debian:buster@sha256:a7789365b226786a0cb9e0f142c515f9f2ede7164a6f6be4a1dc4bfe19d5ec9c
bkrjrzv3nvp7lvzd5cw9vzut7* true 4.879KiB local source for dockerfile
sha256:db193011cbfc238d622d65c4099750758df83d74571e8d7498392b17df381207 true 467.2MiB pulled from docker.io/library/golang:alpine@sha256:a0045fbb52a7ef318937e84cf7ad3301b4d2ba6cecc2d01804f428a1e39d1dfc
wn4m5i5swdcjvt1ud5bvtr75h* true 4.204KiB local source for dockerfile
Reclaimable: 1.08GiB
Total: 1.08GiB
Login to a Registry
$ img login -h
Usage: img login [OPTIONS] [SERVER]
Log in to a Docker registry.
If no server is specified, the default (https://index.docker.io/v1/) is used.
Flags:
-backend backend for snapshots ([fuse naive overlayfs]) (default: naive)
-d enable debug logging (default: false)
-p Password (default: <none>)
-password-stdin Take the password from stdin (default: false)
-state directory to hold the global state (default: /tmp/img)
-u Username (default: <none>)
About
Goals
This project is not trying to reinvent buildkit. If anything all changes and
modifications for buildkit without privileges are being done upstream. The
goal of this project is to in the future just be a glorified cli tool on top of
buildkit.
Snapshotter Backends
naive
The default backend is currently set to naive and if you run it as an
unprivileged user is will not attempt to ever mount.
It can be made unprivileged with mounts and that work is being done, see the unprivileged mounting section below.
overlayfs
You can also use overlayfs
backend, but that requires a kernel patch from Ubuntu to be unprivileged,
see #22.
fuse
The fuse backend runs completely in userspace. It is a bit buggy and a work
in progress so hang tight.
Unprivileged Mounting
To mount a filesystem without root access you need to do it from a mount and user namespace.
Make sure you have user namespace support enabled. On some distros (Debian and
Arch Linux) this requires running echo 1 > /proc/sys/kernel/unprivileged_userns_clone.
You also need a version of runc with the patches from
opencontainers/runc#1688.
Example:
# unshare a mountns and userns
# and remap the user inside the namespaces to your current user
$ unshare -m -U --map-root-user
# then you can run img
$ img build -t user/myimage .
Note that unshare -m -U --map-root-user does not make use of subuid(5)/subgid(5), and also, it disables setgroups(2), which is typically required by apt.
So we might want to use newuidmap(1)/newgidmap(1) SUID binaries to enable these features. See opencontainers/runc#1692 and opencontainers/runc#1693.
If depending on these SUID binaries is problematic, we could use ptrace hacks such as PRoot, although its performance overhead is not negligible. (#15 and AkihiroSuda/runrootless)
For the on-going work toward integrating runc with these patches to buildkit, please refer to moby/buildkit#252
and AkihiroSuda/buildkit_poc@511c7e71.
Contributing
Please do! This is a new project and can use some love <3. Check out the issues.
The local directories are mostly re-implementations of buildkit interfaces to
be unprivileged.
Documentation
¶
There is no documentation for this package.
Source Files
Directories