After reading the .dockerignore
documentation, I'm wondering if there is a way to test it?
Examples
**/node_modules/
How do I check my dockerfile ignore the correct files and directories?
After reading the .dockerignore
documentation, I'm wondering if there is a way to test it?
**/node_modules/
How do I check my dockerfile ignore the correct files and directories?
To expand on Lucas's answer (which implements BMitch's and VonC's answer), you get an easier to understand tree structure with:
$ docker build --no-cache --progress plain --file - . <<EOF
FROM debian
RUN apt update && apt install tree -y
COPY . /build-context
RUN tree /build-context
EOF
which outputs:
#8 0.342 /build-context
#8 0.342 |-- dirA
#8 0.342 | |-- a
#8 0.342 | |-- b
#8 0.342 | | `-- lib
#8 0.342 | | |-- common
#8 0.342 | | | |-- __init__.py
#8 0.342 | | | |-- other.py
...
To expand on Lucas' answer which expands on BMitch's answer (which implements VonC's answer), if you are using buildx you will also need to set the type of progress output by adding --progress plain
and of course to not forget the --no-cache
:
docker build --no-cache --progress plain -f - . <<EOF
FROM busybox
COPY . /build_context
WORKDIR /build_context
RUN find .
EOF
To expand on BMitch's answer (which implements VonC's answer), you can do it with one simple command:
# no cache to ensure output
# progress plain in case you are using buildkit
docker build --no-cache --progress plain --file - . <<EOF
FROM busybox
COPY . /build-context
WORKDIR /build-context
RUN find .
EOF
Switching from CMD
to RUN
means you don't have to ever run the container as the build will output the result of the find .
.
Great solutions,, to add to @Sekenre's answer.
As he mentioned the rsync
matches all files regardless of directory position on the *.bogus
which means if you want to negate the .dockerignore file and ignore everything except for what you want to allow, which results in putting *
at the top of the .dockerignore file, then the rsync method won't transfer any files at all,, even if you have a negated statement for a file just right below it, but a tool to check this would be amazing.
Even having a vscode extension for seeing the files that are ignored and allowed in the file browser would be amazing, just like the git and gitignore thing in vscode file explorer.
To get a detailed analysis of the build context you could use pwaller/docker-show-context.
$ go get -v -u github.com/pwaller/docker-show-context
$ cd ~/path/to/project/using/docker
$ docker-show-context
It outputs statistics about the build such as file sizes and upload times.
To expand on VonC's suggestion, here's a sample build command you can use to create an image with the current folder's build context:
docker image build --no-cache -t build-context -f - . <<EOF
FROM busybox
WORKDIR /build-context
COPY . .
CMD find .
EOF
Once created, run the container and inspect the contents of the /build-context
directory which includes everything not excluded by the .dockerignore
file:
# run the default find command
docker container run --rm build-context
# or inspect it from a shell using
docker container run --rm -it build-context /bin/sh
You can then cleanup with:
docker image rm build-context
Warning: Shameless self-promotion ahead
TL;DR: https://dockerignore.vw.codes
I recently was involved in a project with a convoluted pipeline, and needed rapid prototyping to check the validity of a dockerignore file. I decided to start working on a web utility to quickly be able to verify the output of a dockerignore file.
The secret sauce; validation runs against the actual pattern-matcher function. See Docker's CLI implementation, and here's proof of its usage in my implementation.
It's currently fairly primitive, but I've got plenty of ideas in the pipeline (pun intended). It currently supports:
/api/validate
- docs)I'm planning on adding GitHub authentication for private repo access.
One thing that the other answers do not consider, is that this will potentially copy many gigabytes of data and be very slow, when all you want to do is find out which file(s) you need to exclude to reduce the image size.
So here is how you test your
.dockerignore
without actually copying data:What this will do, is try to sync your current directory with the empty in-memory folder
/dev/shm
verbosely and dry-run (don't actually copy anything) the--exclude-from
option reads glob patterns in the same format as.gitignore
and.dockerignore
You will end up with a list of files copied and a summary with the total size at the end:
Add it to
.dockerignore
:and test again:
This is extremely fast and doesn't fill your disk.
Edit: There is one difference that I have found. For rsync the pattern
*.bogus
matches all files with that name regardless of the directory..dockerignore
however only matches*.bogus
in the current directory. To get the same behavior you need to prefix the pattern with the path glob characters**/*.bogus
This will still work with rsync.