alexpires.me/a13labs.infra
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity
Files
350650ecc288e7ceee8119d98def36f80a34d580
a13labs.infra/ansible/host_vars/gpu-01.lab.alexpires.me.yml
T
alexandre.pires 350650ecc2 Add GPU monitoring services and configurations 
- Create systemd service templates for llama_exporter, nvidia_exporter, and podman.
- Add Prometheus configuration template for GPU metrics scraping.
- Introduce variables for GPU monitoring role in main.yml.
- Implement tasks for syncing llama models, including user setup and package installation.
- Update podman tasks to ensure proper sudo access and lingering for the podman user.
- Modify inventory to include gpu_monitoring group.
- Add Terraform modules for deploying Vaultwarden, including ingress and service configurations.
- Create Grafana dashboard for real-time GPU monitoring metrics.
- Update secrets and environment files to include Vaultwarden admin token.
2026-05-29 23:42:14 +02:00

195 lines
6.9 KiB
YAML
Raw Blame History

hostname: gpu-01.lab.alexpires.me
ansible_user: "{{ lookup('env', 'ANSIBLE_USER') }}"
network_interface: enp7s0
# Users
login_users:
- username: "{{ ansible_user }}"
comment: "Managed by Ansible"
sudoer: true
sudoer_root_only: true
sudoer_no_password: true
pubkey: "{{ lookup('file', 'keys/ansible_user.pub') }}"
password_disabled: true
- username: operator
comment: "Managed by Ansible"
sudoer: true
pubkey: "{{ lookup('file', 'keys/operator.pub') }}"
password_expiration: false
# SSH Connection and security
sshd_port: 22
ssh_allowed_networks:
- "10.5.5.5/32" # Wireguard VPN (Laptop)
- "10.19.4.0/24" # local network
sshd_admin_net: "10.19.4.0/24" # local network
sshd_allow_groups: "{{ ansible_user }} users"
sshd_permit_root_login: "no"
sshd_password_authentication: "no"
sshd_allow_tcp_forwarding: "yes"
# Due to an issue increase the number of max auth tries
sshd_max_auth_tries: 10
packages_blocklist: []
# Firewall ports
ufw_enable: true
fw_allowed_ports:
- { rule: "allow", port: "22", proto: "tcp", from: "10.5.5.5/32" } # Wireguard VPN (Laptop)
- { rule: "allow", port: "22", proto: "tcp", from: "10.19.4.0/24" } # local network
- { rule: "allow", port: "11434", proto: "tcp", from: "10.19.4.0/24" } # Ollama (local network)
- { rule: "allow", port: "11434", proto: "tcp", from: "10.5.5.5/32" } # Ollama (Laptop)
- { rule: "allow", port: "8880", proto: "tcp", from: "10.19.4.0/24" } # Kokoro-FastAPI (local network)
- { rule: "allow", port: "8880", proto: "tcp", from: "10.5.5.5/32" } # Kokoro-FastAPI (Laptop)
- { rule: "allow", port: "8188", proto: "tcp", from: "10.19.4.0/24" } # ComfyUI (local network)
- { rule: "allow", port: "8999", proto: "tcp", from: "10.19.4.0/24" } # ComfyUI - SimpleHttpServer (local network)
- { rule: "allow", port: "8012", proto: "tcp", from: "10.19.4.0/24" } # Qwen2 (local network)
- { rule: "allow", port: "8012", proto: "tcp", from: "10.5.5.5/32" } # Qwen2 (Laptop)
- { rule: "allow", port: "9091", proto: "tcp", from: "10.19.4.0/24" } # Prometheus (local network)
- { rule: "allow", port: "9091", proto: "tcp", from: "10.5.5.5/32" } # Prometheus (Laptop)
ufw_outgoing_traffic:
- 22 # SSH
- 53 # DNS
- 80 # HTTP
- 123 # NTP
- 443 # HTTPS
- 853 # DNS over TLS
- 139 # Samba
- 445 # Samba
ufw_default_forward_policy: "ACCEPT"
fw_loopback_traffic_deny: false
ipv4_sysctl_settings:
net.ipv4.ip_forward: 1
disable_ipv6: true
# geoip will override hosts_allow and hosts_deny (so it's disabled - CIS hardening)
custom_hosts_acls: true
# This fixes polkit issues with sudoers file (CIS hardening)
hide_pid: 0
# CIS by default changes compilers permissions, Nvidia KMod needs to be build
compilers:
# aide
install_aide: false
# Unattended upgrades
unattended_reboot: true
unattended_reboot_time: "04:30"
# Duo Security
duo_users: ["*", "!provision", "!comfyui", "!kokoro", "!ollama"]
duo_api_hostname: api-7cda1654.duosecurity.com
# podman services settings
ollama_home: "/mnt/data/ollama"
llama_home: "/mnt/data/llamacpp"
podman_user_pubkey: "ecdsa-sha2-nistp521 AAAAE2VjZHNhLXNoYTItbmlzdHA1MjEAAAAIbmlzdHA1MjEAAACFBAHHOPBR9p9kq5Cqzpe4cr3jHnweaYrHPXv5sXNzt+sCXP54uc5rWUBhxW2OQVvQzJ47dEVhEKi4WSC7LcuKS2G5AQDzWXNiasHvYIYQU3F/EknVCZnsiXYqXphYkJA6rJCNRnISZCIC1mocq6PB7J08ONdRFCvjfUBuVRT8BNGXNmQ/zQ=="
# ComfyUI settings
comfyui_home: "/mnt/data/comfyui"
comfyui_user_pubkey: "ecdsa-sha2-nistp521 AAAAE2VjZHNhLXNoYTItbmlzdHA1MjEAAAAIbmlzdHA1MjEAAACFBAHHOPBR9p9kq5Cqzpe4cr3jHnweaYrHPXv5sXNzt+sCXP54uc5rWUBhxW2OQVvQzJ47dEVhEKi4WSC7LcuKS2G5AQDzWXNiasHvYIYQU3F/EknVCZnsiXYqXphYkJA6rJCNRnISZCIC1mocq6PB7J08ONdRFCvjfUBuVRT8BNGXNmQ/zQ=="
# Kokoro FastAPI settings (we don't set home because it is not expected to use a lot a disk space)
kokoro_user_pubkey: "ecdsa-sha2-nistp521 AAAAE2VjZHNhLXNoYTItbmlzdHA1MjEAAAAIbmlzdHA1MjEAAACFBAHHOPBR9p9kq5Cqzpe4cr3jHnweaYrHPXv5sXNzt+sCXP54uc5rWUBhxW2OQVvQzJ47dEVhEKi4WSC7LcuKS2G5AQDzWXNiasHvYIYQU3F/EknVCZnsiXYqXphYkJA6rJCNRnISZCIC1mocq6PB7J08ONdRFCvjfUBuVRT8BNGXNmQ/zQ=="
# Let's Encrypt / Certbot
cloudns_auth_id: "{{ lookup('env', 'CLOUDNS_AUTH_ID') }}"
cloudns_auth_password: "{{ lookup('env', 'CLOUDNS_PASSWORD') }}"
cloudns_nameserver: "109.201.133.111"
certbot_email: c.alexandre.pires@alexpires.me
certbot_domains:
- "gpu-01.lab.alexpires.me"
certbot_cockpit_domain: "gpu-01.lab.alexpires.me"
ollama_models:
# Thinking models
- gpt-oss:latest
- gemma4:e4b
- qwen3.5:9b
- qwen3.5:4b
# Non-thinking models
- ministral-3:3b
- ministral-3:8b
- ministral-3:14b
- granite4.1:3b
- granite4.1:8b
- deepseek-r1:latest
- qwen2.5-coder:3b
# Embedding models
- embeddinggemma:latest
- nomic-embed-text:latest
# Whisper models
- dimavz/whisper-tiny:latest
ollama_model_files:
- file: "ollama/gpt-oss-latest-32k.modelfile"
name: "gpt-oss-32k:latest"
- file: "ollama/gemma4-e4b-32k.modelfile"
name: "gemma4-e4b-32k:latest"
- file: "ollama/qwen3.5-9b-32k.modelfile"
name: "qwen3.5-9b-32k:latest"
- file: "ollama/qwen3.5-9b-64k.modelfile"
name: "qwen3.5-9b-64k:latest"
- file: "ollama/qwen3.5-4b-32k.modelfile"
name: "qwen3.5-4b-32k:latest"
- file: "ollama/qwen3.5-4b-64k.modelfile"
name: "qwen3.5-4b-64k:latest"
llama_models:
# TODO: replace revision values with pinned commit SHAs for strict reproducibility.
- name: "gpt-oss-20b-q4-0"
model_id: "unsloth/gpt-oss-20b-GGUF"
quant: "Q4_0"
revision: "main"
- name: "gemma-4-26b-a4b-ud-iq4-xs"
model_id: "unsloth/gemma-4-26B-A4B-it-GGUF"
quant: "UD-IQ4_XS"
revision: "main"
- name: "gemma-4-31b-iq4-xs"
model_id: "unsloth/gemma-4-31B-it-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "qwen3-6-35b-a3b-mtp-ud-iq4-xs"
model_id: "unsloth/Qwen3.6-35B-A3B-MTP-GGUF"
quant: "UD-IQ4_XS"
revision: "main"
- name: "qwen3-5-27b-mtp-iq4-xs"
model_id: "unsloth/Qwen3.5-27B-MTP-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "qwen3-5-35b-a3b-ud-iq4-xs"
model_id: "unsloth/Qwen3.5-35B-A3B-GGUF"
quant: "UD-IQ4_XS"
revision: "main"
- name: "ministral-3-14b-instruct-iq4-xs"
model_id: "unsloth/Ministral-3-14B-Instruct-2512-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "granite-4-1-30b-iq4-xs"
model_id: "unsloth/granite-4.1-30b-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "qwen3-5-9b-iq4-xs"
model_id: "unsloth/Qwen3.5-9B-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "qwen3-6-35b-a3b-claude47-distilled-iq4-xs"
model_id: "lordx64/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled-IQ4_XS-GGUF"
quant: "IQ4_XS"
revision: "main"
- name: "devstral-small-2-24b-instruct-iq4-xs"
model_id: "unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF"
quant: "IQ4_XS"
revision: "main"
# Optional global model preset options applied under [*] in models.ini.
# Keys map to llama.cpp CLI args without leading dashes, short args, or LLAMA_ARG_* env names.
llama_preset_global:
c: 131072
n-gpu-layers: all
Reference in New Issue View Git Blame Copy Permalink