Upgrade LlamaSrv
Use upgrade.sh for an existing LlamaSrv installation. It updates the application, Python dependencies, and service definitions while retaining your configuration.
From the repository checkout on the installed host:
git pull --ff-only
./scripts/upgrade.sh --check
sudo ./scripts/upgrade.sh
The upgrade validates settings and model files, updates the installation, and restarts inference and HTTP without a health probe. It records the new version after deployment checks and both service start commands succeed. Inference may still be loading its model when the upgrade completes.
Your /opt/prod/llamasrv/conf/llamasrv.json, models, and external llama.cpp installation are preserved. There is no automatic rollback: if an upgrade fails, review the error and logs before retrying.
sudo journalctl -u llamasrv-llm.service -u llamasrv-http.service -n 100 --no-pager
curl --fail http://127.0.0.1:51000/health
For a new host, start with Install LlamaSrv.