boinc-server-docker is the easiest way to run your own BOINC server. You can run the server on a Linux machine, in which case the requirements are,
- Docker (>=17.09.0ce)
- docker-compose (>=1.17.0 but !=1.19.0 due to a bug)
- git
or you can run your server on Windows 7+ or Mac OSX, in which case you should use either,
- Docker for Mac (>=17.09.0ce)
- Docker for Windows (>=17.09.0ce)
or, if your Windows/Mac system is too old to support either of those,
- Docker Toolbox (>=17.09.0ce)
There are no other dependencies, as everything else is packaged inside of Docker.
For a full tutorial on creating your own server with boinc-server-docker, see the project cookbook.
If you are looking to create a server and are already somewhat familiar with Docker and BOINC, the following short description takes you through creating a server and running your own science application.
To check out this repository and get a test server fully up and running, simply run,
git clone https://github.com/marius311/boinc-server-docker.git
cd boinc-server-docker
docker-compose pull
docker-compose up -dYou can now visit the server webpage and connect clients to the server at http://127.0.0.1/boincserver.
Notes:
- The first time you run this, it may take a few minutes after invoking the
docker-compose up -dcommand before the server webpage appears.- Make sure your user is added to the
dockergroup, otherwise thedocker-composeanddockercommands in this guide need to be run withsudo.- If using Docker Toolbox, replace the final command above with
URL_BASE=$(docker-machine ip) docker-compose up -d. The server will be accessible at the IP returned bydocker-machine iprather than at127.0.0.1.
The server comes pre-configured to immediately run Docker-based science applications. To do so, first create a Docker container which runs your code. As an example, we will use a python:alpine image. Suppose your calculation is run with the following command,
docker run python:alpine python -c "print('Hello BOINC')"To submit a job on the server which runs this as a BOINC job you would first get a shell inside the server,
docker-compose exec apache bashThen submit the job by running
bin/boinc2docker_create_work.py python:alpine python -c "print('Hello BOINC')"Now you can connect a BOINC client the server and run this job. Note that to run these types of Docker-based jobs, the client computer will need 64bit Virtualbox installed and "virtualization" enabled in the BIOS.
If your job has output files, have the container write them in /root/shared/results and they are automatically returned to the server when the job is done.
This is a simple example, but any Docker containers with arbitrary code installed inside of them will work!
To stop the server and delete all server and database files (for example, if you want to start over with a fresh copy), run,
docker-compose down -vFinally, boinc-server-docker is not just useful to get a simple test server running, its actually meant to run your real server. To learn how to, read the project cookbook, or look at the Cosmology@Home source code as an example (boinc-server-docker was in fact originally developed for Cosmology@Home).
Happy crunching!
If you wish to modify and rebuild any of the boinc-server-docker images yourself, you will need this git repository's submodules checked out. To do so, run git submodule update --init --recursive from the repository folder, or clone with git clone --recursive in the first place. Note that building these images is only necessary if you are helping with development of this package; if you wish to build your own project using these base images, follow the instruction in the cookbook instead.
Currently, building the images is only guaranteed to work on Linux. Some users have reported successfully building on Windows or Mac, but this is considered experimental at this point.
Please feel free to contact the maintainers or submit Issues and Pull Requests for this repository if you wish to contribute!
-
Version 4.1.0 - Oct 20, 2019
- Based on server_release/1.2/1.2.
-
Version 4.0.2 - Apr 27, 2019
- Based on server_release/1.0/1.0.4.
-
Version 4.0.1 - Jan 31, 2018
- Fix problem with mysql image-tagging which caused errors when trying to build a custom project.
-
Version 4.0.0 - Jan 18, 2018
- Based on server_release/1.0/1.0.3.
-
Version 3.0.1 - Aug 2, 2018
- Minor bug fix where
PROJECT_ROOTwasn't fully customizable
- Minor bug fix where
-
Version 3.0.0 - July 27, 2018
- Based on server_release/0.9.
- Upgraded to Debian Stretch, PHP 7.0.31 and MariaDB 10.3.8.
- Docker requirement is now 17.09.0ce on all platforms.
- Project "secrets" such as passwords and signing keys are now stored in a new volume called
secrets, and the procedure for how to deal with them is documented here. - Breaking change: For improved security, the BOINC daemons no longer run as root, instead they run as an unprivileged user, by default named
boincadm. - Added two new options which are configurable at build-time,
BOINC_USERandPROJECT_ROOT, and fixedPROJECTwhich wasn't fully configurable before. Under the hood, theboinc-server-dockerimages now use DockerONBUILDinstructions to make this happen. - Upgrade instructions: If you don't care about the files in your project's database and project folder, you can just wipe your project clean with
docker-compose down -vand simply start a fresh server with version3.0.0. If instead you want to upgrade a project you created withboinc-server-docker v2.x.x, you should follow these instructions:-
Edit the
FROMline in your custom Dockerfiles to source the appropriate3.0.0images. -
Diff your
docker-compose.ymland.envfiles against the corresponding ones inexample_project/, and merge in changes you see (notably, add thesecretsvolume). -
Run
docker-compose buildto build updated images. -
Run
docker-compose run --rm makeproject bashand navigate to/home/boincadm/secrets. This is yoursecretsvolume, and you should edit the files you see here so that they contain your passwords, keys, etc... -
Bring your project down with
docker-compose down. -
Run the following to update permissions and upgrade your database:
source .env eval "$(docker-compose run --rm -T makeproject cat /run/secrets/secrets.env)" docker-compose run --rm -u root makeproject chown -R $BOINC_USER:$BOINC_USER $PROJECT_ROOT.dst docker-compose exec mysql mysql_upgrade docker-compose exec mysql mysqladmin -u root password $DB_PASSWD
-
Now bring your project back up with
docker-compose up -d.
-
-
Version 2.1.0 - May 29, 2018
- Update boinc to server_release/0.9.
-
Version 2.0.0 - Feb 27, 2018
- New feature: The server URL and project name can now be changed at run-time with e.g.:
URL_BASE=http//1.2.3.4 PROJECT=myproject docker-compose up -d. See here and here in the Project Cookbook for more details. - Breaking change: The
$PROJHOMEvariable which was previously available inapacheandmakeprojectcontainers has been renamed to$PROJECT_ROOTto be consistent with themake_toolsscript, similarly as withURL_BASEandPROJECT, and in anticipation that it too will become configurable. - Upgraded version requirements.
- New feature: The server URL and project name can now be changed at run-time with e.g.:
-
Version 1.4.1 - July 26, 2017
- The default server URL is now
http://127.0.0.1/boincserverrather than previously when it washttp://boincserver.com/boincserver. This removes the need to edit your/etc/hostsfile on Linux, and on Windows/Mac/docker-machine replaces having to edit/etc/hostswith the SSH tunnel command above. Related warning: the boincserver.com domain is currently being squatted, so if you're using the old version be careful that you do not type sensitive information into the server website thinking you're interacting with your local test server when in fact it's a remote server at the squatted domain. - Updated docker-compose requirement from 1.6.0 to 1.7.0, and on Windows/Mac, updated Docker Toolbox requirement from 1.10.0 to 1.11.0
- A number of improvements to boinc2docker (see ccfe9a9).
- The default server URL is now