[Unit]
Description=Podman file for llama.cpp server
After=local-fs.target network-online.target nvidia-cdi-generator.service

[Service]
Type=oneshot
User={{ llama_server_user }}
Group={{ llama_server_group }}
RemainAfterExit=true
ExecStart={{ podman_binary }} start llamacpp
ExecStop={{ podman_binary }} stop llamacpp

StandardOutput=journal
StandardError=journal

[Install]
WantedBy=multi-user.target