Files
turboquant/ansible/deploy_turboquant.yml
STEP35 CLI e20439b544
All checks were successful
Smoke Test / smoke (pull_request) Successful in 7s
deploy: Ansible role for TurboQuant-compressed Gemma 4 across fleet nodes (#98)
- Adds ansible/ deploy_turboquant.yml playbook with per-node config
- Adds turboquant-deploy role: OS-specific (darwin/debian) tasks
- Adds health_check.sh and integration test (chat completion)
- Adds inventory.ini.example with Mac/Allegro/Ezra groups
- Deploys llama.cpp with TurboQuant (Metal on macOS)
- Systemd service (Linux) with TURBO_LAYER_ADAPTIVE env
2026-04-26 06:55:35 -04:00

70 lines
1.8 KiB
YAML

---
# deploy_turboquant.yml — Deploy TurboQuant across fleet nodes
# Usage: ansible-playbook -i ansible/inventory.ini ansible/deploy_turboquant.yml
- name: Deploy TurboQuant to Mac (local)
hosts: mac
become: yes
gather_facts: yes
vars:
turboquant_user: "turboquant"
turboquant_install_dir: "/opt/turboquant"
turboquant_service_name: "turboquant"
turboquant_port: 8081
turboquant_host: "0.0.0.0"
turboquant_context: 131072
turboquant_model: "gemma-4"
turboquant_model_file: "gemma-4-26B-A4B.gguf"
turboquant_kv_type: "turbo4"
turboquant_layer_adaptive: 7
node_preset: "turboquant_k8v4"
node_hardware: "M1-16GB"
roles:
- turboquant-deploy
- name: Deploy TurboQuant to Allegro VPS
hosts: allegro
become: yes
gather_facts: yes
vars:
turboquant_user: "turboquant"
turboquant_install_dir: "/opt/turboquant"
turboquant_service_name: "turboquant"
turboquant_port: 8081
turboquant_host: "0.0.0.0"
turboquant_context: 65536
turboquant_model: "gemma-4-E4B"
turboquant_model_file: "gemma-4-E4B.gguf"
turboquant_kv_type: "q4_0"
turboquant_layer_adaptive: 0
node_preset: "turboquant_4bit_nc"
node_hardware: "VPS-2c8g"
roles:
- turboquant-deploy
- name: Deploy TurboQuant to Ezra VPS
hosts: ezra
become: yes
gather_facts: yes
vars:
turboquant_user: "turboquant"
turboquant_install_dir: "/opt/turboquant"
turboquant_service_name: "turboquant"
turboquant_port: 8081
turboquant_host: "0.0.0.0"
turboquant_context: 65536
turboquant_model: "gemma-4-E4B"
turboquant_model_file: "gemma-4-E4B.gguf"
turboquant_kv_type: "q4_0"
turboquant_layer_adaptive: 0
node_preset: "turboquant_4bit_nc"
node_hardware: "VPS-2c8g"
roles:
- turboquant-deploy