Closed
Description
Now FreeBSD testing looks so: we have a Linux machine (hardware, non-virtual) with GitHub Runner, which serves jobs with freebsd-sh
label. Internally it starts a VirtualBox machine with FreeBSD and and run testing inside it bt ssh.
I propose the following scheme:
- Prepare FreeBSD image for MCS.
- Consult with MCS guys and provide it as one of available images for everyone?
- Investigate how to setup GitHub Runner inside.
- Run several virtual machines with FreeBSD and GitHub Runner.
- Revisit
.workflows
/.gitlab.mk
/.travis.mk
part in the repository. - Don't forget to update internal machines list.
After this we can at least scale those runners manually. In future, I hope, I'll implement some autoscaling and, what is even more important, running a job of always-clean virtual machine.
There is Terraform machinery for manage self-hosted GitHub Runners (this part of the infrastructure is private, keyword: mfactory). Sooner or later, I hope, we'll track and manage our runners this way.