download-lb-cert:
Downloads cert and key files provided by the load balancer. It atomically
makes them available at /ie/workdir/ssl.
setup-replica/replica_setup.rb:
Creates these systemd units:
- inference-engine.service: Runs an inference engine, e.g. vllm.
- inference-gateway.service: Runs our inference gateway. It receives requests
through the load balancer and forwards them to the inference engine. It
performs TLS termination, checks API keys, ...
- lb-cert-download.service: Downloads the cert and key files from the load
balancer and then exits.
- lb-cert-download.timer: Triggers lb-cert-download.service.