This is a "proof of concept" to enble processing and testing of AnyBody models on freely available compute resources in the cloud. This repository contains a simple toy model, which is automatically run/tested every time a new commit is pushed to the repository.
This workflow uses GitHub actions to run AnyBody on GitHub's linux machines in the cloud. The only requirement is a 'floating' license for AnyBody.
Note: This is a work in progress. If you happen to need this kind of setup, please contact us on [email protected]
The cloud workflow is defined in the .github/workflows/anybodycon.yml file.
To use AnyBody in GitHub actions you must specify our linux docker container: with AnyBody preinstalled:
jobs:
job-with-anybody:
runs-on: ubuntu-latest
container: ghcr.io/anybody/anybodycon-github-actions:latest
... AnyBody can only run if you have a valid license. It requires a floating license and license server which is exposed to the internet.
The license server's port and address is specified with an environment variable in the action yaml file:
env:
RLM_LICENSE: <port>@<adress>When the license server is publicly available on the internet, it must be secured with a password.
The license server password is also defined with an environment variable (RLM_LICENSE_PASSWORD), but to prevent anyone from obtaining the password it should be stored with GitHub's system for encrypted screets.
With that in place the license server password can then be specified in the steps that launch AnyBody:
- name: Run AnyBodyCon
env:
RLM_LICENSE_PASSWORD: ${{ secrets.LICENSE_PASSWORD }}
run: |
anybodycon -m macrofile.anymcr
Github actions is free for public repositories, and has a free teir for private repositories. But the purpose of this setup is not to process very large datasets. But i allows you to run normal models and setup automatic tests of your models which are hosted on GitHub.