Closed rtyler closed 3 years ago
My initial thoughts around this was that I would just create an agent work directory and then put work/<pipeline-uuid>
inside of it for each pipeline that the agent runs.
This runs into some problems insofar that each pipeline run would need to clone from scratch if running on the same agent. :thinking:
The agent doesn't currently create any working directories for itself, and needs to . Each pipeline needs to be run in its own space.
I'm thinking the agent CLI needs to be invoked with a certain base directory, and then the agent can handle building pipeline-specific directories off of that.