…~arl/elkweb-docker

updated names

Using docker-compose's feature to check wether the server is still
running from inside the container.

Since the process inside the container eats up all the server resources
and we don't know why, limiting the available resources to mitigate the
problem seems a legit idea.
The container might use half a CPU and 512 MB of memory.

Avoiding local cache data as well.

This change was in use on production server before my time, just
committing changes to repo here.

There was a hardcoded IP address of the production server webapp.rtsys
in port binding configuration.
There's no need for that and it blocks using the docker-compose setup
locally. Thus, it's removed and we expose the block to all interfaces,
via 0.0.0.0.

…admin/elkweb-docker into sho/docker

This commit merges Docker related files from separate repository into
elklive, providing a runtime/development enviroment. It could be used to
build and distribute images as well.
This commit merges in history from Docker files which goes back to
before my time at the working group, but has some addition from me as
well.

Copying project sources into Docker container image from working copy
instead of cloning it again.
Cloning sources inside the container is cached by Docker image layer
and, thus, never get updated during rebuild. Some build steps rely on
git commands which is why .git/ is not ignored in Docker COPY.
Also, we are stuck to Alpine v3.19 due to Java incompatibilities with
Gradle 8 or newer. Hard coding alpine:3.19 image to work around this
issue for now.

Docker compose version is obsolete.
https://github.com/compose-spec/compose-spec/blob/master/spec.md#version-and-name-top-level-elements

No need to copy docker-compose.yml into Docker image.

There was a git command-line call when yarn builds webpack components of
the client.
This fails in some build environemnts (e.g. Github build actions or
standard Docker builds) when no .git/ folder is found in project root.

We already have a version information declared in package.json for NPM
description. Use it to render in web pages instead of hard coded string
or plain git commit hash.
Later on, the version information might get bumped when building
releases and reused to set tags in git and to Docker images.

When a commit hash is provided via GitHub's build action environment
variable it's concatinated to the version string rendered to web pages.

It's easier to read.

Group ID of non-privileged user can now also be passed from outside via
Docker argument. This is beneficial when binding your work directory to
actively develop the app using a running container instance.
By setting matching file permission inside container environment you
avoid mixing up file permission between both sides of the mounted
folder.

Defaulting to dynamically assigned host port which eases starting
multiple container (instances).
Add an environment option to specify host port explicitly, e.g. for
production environment.

Forcing the user to copy the file and activate relevant lines by
removing leading hash character.

Some build time dependency caches that should be ignored from version
control and from being copied back into container image as well.

Comment out volumes section to leave it as option for local development.

This is the default UID:GID mapping now as the previously hard-coded
values of 1002:1002 where used in production system, only.

When building locally a 'latest' tag is added to each new image. As it
seems, this is done automatically only for tagged releases in GitHub
Actions.
Let's try setting this tag explicitly as it's the default tag for a
image specification when no other is given.

This reverts commit 6702560.