diff options
Diffstat (limited to 'reproduce/software/containers/README-docker.md')
| -rw-r--r-- | reproduce/software/containers/README-docker.md | 43 |
1 files changed, 32 insertions, 11 deletions
diff --git a/reproduce/software/containers/README-docker.md b/reproduce/software/containers/README-docker.md index f86dceb..d651e22 100644 --- a/reproduce/software/containers/README-docker.md +++ b/reproduce/software/containers/README-docker.md @@ -35,17 +35,20 @@ software environment) to easily move it from one computer to another. systemctl start docker ``` - 2. Using your favorite text editor, create a `docker-local.sh` in your top - Maneage directory (as described in the comments at the start of the - `docker.sh` script in this directory). Just activate `--build-only` on - the first run so it doesn't go onto doing the analysis and just sets up - the software environment. - - 3. After the setup is complete, run the following command to confirm that - the `maneage-base` (the OS of the container) and `maneaged` (your - project's full Maneage'd environment) images are available. If you want - different names for these images, add the `--project-name` and - `--base-name` options to the `docker.sh` call. + 2. Using your favorite text editor, create a `run.sh` in your top Maneage + directory (as described in the comments at the start of the `docker.sh` + script in this directory). Just activate `--build-only` on the first + run so it doesn't go onto doing the analysis and just sets up the + software environment. Set the respective directory(s) based on your + filesystem (the software directory is optional). The `run.sh` file name + is already in `.gitignore` (because it contains local directories), so + Git will ignore it and it won't be committed by mistake. + + 3. After the setup is complete, remove the `--build-only` and run the + command below to confirm that `maneage-base` (the OS of the container) + and `maneaged` (your project's full Maneage'd environment) images are + available. If you want different names for these images, add the + `--project-name` and `--base-name` options to the `docker.sh` call. ```shell docker image list @@ -85,6 +88,24 @@ image into it. Below are some useful Docker usage scenarios that have proved to be relevant for us in Maneage'd projects. +### Saving and loading an image as a file + +Docker keeps its images in hard to access (by humans) location on the +operating system. Very much like Git, but with much less elegance: the +place is shared by all users and projects of the system. So they are not +easy to archive for usage on another system at a low-level. But it does +have an interface (`docker save`) to copy all the relevant files within an +image into a tar ball that you can archive externally. There is also a +separate interface to load the tarball back into docker (`docker load`). + +Both of these have been implemented as the `--image-file` option of the +`docker.sh` script. If you want to save your Maneage'd image into an image, +simply give the tarball name to this option. Alternatively, if you already +have a tarball and want to load it into Docker, give it to this option once +(until you "clean up", as explained below). In fact, docker images take a +lot of space and it is better to "clean up" regularly. And the only way you +can clean up safely is through saving your needed images as a file. + ### Cleaning up Docker has stored many large files in your operating system that can drain |