Something I believe this project could do well is concurrent operations for install. The compute is assumed to exist while you are performing the install. This means we can:
Concurrently copy artifacts to all nodes (safely)
Begin installation of the primary server node
While the primary server node is installing - ensure all other nodes are bootstapped to the point where the systemd service just needs to be started
Secondary server nodes should receive a signal that. primary is ready - install all
Agent nodees should receive a signal when server nodes are complete / healthy - install all
Instead of the serial process of installation - bootstrap and install concurrently as much as possible.
Something I believe this project could do well is concurrent operations for install. The compute is assumed to exist while you are performing the install. This means we can:
Instead of the serial process of installation - bootstrap and install concurrently as much as possible.