feat(vaultwarden): add SMTP configuration options and enhance signup settings

- Introduced SMTP settings for Vaultwarden including host, port, security, and authentication details. - Added variables for signup verification, 2FA settings, password hints, and logging options. - Updated Vaultwarden deployment to utilize new SMTP configurations. - Enhanced Grafana module to support dynamic dashboard and datasource provisioning. - Added LLM proxy configuration for Open Web UI with necessary environment variables.
2026-05-30 23:24:44 +02:00
parent 350650ecc2
commit ec740f458f
89 changed files with 3154 additions and 856 deletions
@@ -46,18 +46,6 @@ amd_exporter_python_version: "python3"
 amd_exporter_sysfs_gpu_pattern: "/sys/class/drm/card?/device/"

 # Llama Exporter configuration
-llama_exporter_enabled: true
-llama_exporter_port: 9550
-llama_exporter_user: llama_exporter
-llama_exporter_user_home: "/home/llama_exporter"
-llama_exporter_install_dir: "/opt/llama_exporter"
-llama_exporter_venv_path: "/opt/llama_exporter/.venv"
-llama_exporter_scrape_interval: 15
-llama_exporter_scrape_timeout: 5
-
-# Llama.cpp scrape targets
-llama_exporter_targets: []
-
 # Firewall configuration
 prometheus_fw_port: "{{ prometheus_port }}"
 prometheus_fw_proto: "tcp"
@@ -1,176 +0,0 @@
-#!/usr/bin/env python3
-"""LLM inference metrics exporter for Prometheus.
-
-Scrape targets:
-    - llama.cpp HTTP API servers (standard /metrics or custom health)
-"""
-
-import json
-import os
-import time
-import logging
-import urllib.request
-import urllib.error
-from http.server import HTTPServer, BaseHTTPRequestHandler
-
-logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
-logger = logging.getLogger(__name__)
-
-POLL_INTERVAL = int(os.environ.get("LLAMA_EXPORTER_INTERVAL", "15"))
-SCRAPE_TIMEOUT = int(os.environ.get("LLAMA_EXPORTER_TIMEOUT", "5"))
-
-LLAMA_CPP_DEFAULTS = [
-    {"name": "llama-cpp-8012", "url": "http://localhost:8012"},
-]
-
-
-def _fetch_json(url, timeout=SCRAPE_TIMEOUT):
-    try:
-        req = urllib.request.Request(url, headers={"Accept": "application/json"})
-        with urllib.request.urlopen(req, timeout=timeout) as resp:
-            return json.loads(resp.read().decode("utf-8"))
-    except Exception as e:
-        logger.debug("Failed to fetch %s: %s", url, e)
-        return None
-
-
-def _collect_llama_cpp(target):
-    metrics = {}
-
-    def _set_metric(metric_name, labels, value):
-        label_key = tuple(sorted(labels.items()))
-        metrics[(metric_name, label_key)] = value
-
-    base_url = target["url"].rstrip("/")
-
-    # Check health endpoint
-    health_url = f"{base_url}/health"
-    health = _fetch_json(health_url)
-    if health and isinstance(health, dict):
-        status = health.get("status", "unknown")
-        _set_metric("llama_server_health", {"server": target["name"], "model": health.get("model", "unknown")}, 1.0 if status == "ok" else 0.0)
-    else:
-        _set_metric("llama_server_health", {"server": target["name"], "model": "unknown"}, 0.0)
-
-    # Check /metrics endpoint for llama.cpp built-in metrics
-    metrics_url = f"{base_url}/metrics"
-    try:
-        req = urllib.request.Request(metrics_url, headers={"Accept": "text/plain"})
-        with urllib.request.urlopen(req, timeout=SCRAPE_TIMEOUT) as resp:
-            body = resp.read().decode("utf-8")
-            for line in body.splitlines():
-                line = line.strip()
-                if not line or line.startswith("#"):
-                    continue
-                # Parse simple key value pairs
-                if "{" in line:
-                    key_part = line.split("{")[0]
-                    labels_part = line.split("{")[1].rstrip("}").split("}")[0] if "}" in line else ""
-                    value = line.split()[-1] if line.split() else "1"
-                    label_str = ""
-                    if labels_part:
-                        label_str = ",".join(f'{{{k}="{v}"}}' for k, v in (
-                            pair.split("=") for pair in labels_part.split(",") if "=" in pair
-                        ))
-                    _set_metric(key_part.strip(), {"server": target["name"], "raw": labels_part}, float(value) if value.replace(".", "").replace("-", "").isdigit() else value)
-                else:
-                    parts = line.split()
-                    if len(parts) >= 2:
-                        try:
-                            float(parts[1])
-                            _set_metric(parts[0], {"server": target["name"]}, parts[1])
-                        except ValueError:
-                            pass
-    except Exception as e:
-        logger.debug("Failed to scrape %s/metrics: %s", metrics_url, e)
-
-    # Check /model endpoint for model info
-    model_url = f"{base_url}/models"
-    models = _fetch_json(model_url)
-    if models and isinstance(models, dict):
-        model_list = models.get("data", [models]) if "data" in models else [models]
-        for m in model_list:
-            if not isinstance(m, dict):
-                continue
-            model_id = m.get("id", "unknown")
-            _set_metric("llama_models_loaded", {"server": target["name"], "model": model_id}, 1.0)
-            ctx_max = m.get("context_size", m.get("n_ctx", 0))
-            if ctx_max:
-                _set_metric("llama_model_context_max", {"server": target["name"], "model": model_id}, int(ctx_max))
-            trained = m.get("train_tokens", m.get("n_train", 0))
-            if trained:
-                _set_metric("llama_model_tokens_trained", {"server": target["name"], "model": model_id}, int(trained))
-
-    return metrics
-
-
-
-class MetricsHandler(BaseHTTPRequestHandler):
-    def do_GET(self):
-        if self.path != "/metrics":
-            self.send_response(404)
-            self.end_headers()
-            return
-
-        all_metrics = {}
-
-        # Scrape llama.cpp targets
-        targets = []
-        env_targets = os.environ.get("LLAMA_TARGETS", "")
-        if env_targets:
-            try:
-                targets = json.loads(env_targets)
-            except json.JSONDecodeError:
-                logger.warning("Invalid LLAMA_TARGETS JSON, using defaults")
-                targets = []
-
-        if not targets:
-            targets = LLAMA_CPP_DEFAULTS
-
-        for target in targets:
-            if "url" not in target or "name" not in target:
-                target["name"] = target.get("name", "default")
-                target["url"] = target.get("url", "http://localhost:8012")
-            try:
-                all_metrics.update(_collect_llama_cpp(target))
-            except Exception as e:
-                logger.error("Error scraping llama.cpp target %s: %s", target.get("name", "unknown"), e)
-
-        lines = []
-        for (metric_type, label_pairs), value in sorted(all_metrics.items()):
-            labels = dict(label_pairs)
-            if not labels:
-                line = f"{metric_type} {value}"
-            else:
-                label_str = ",".join(f'{k}="{v}"' for k, v in sorted(labels.items()))
-                line = f"{metric_type}{{{label_str}}} {value}"
-            lines.append(line)
-
-        body = "\n".join(lines) + "\n" if lines else "# no metrics available\n"
-
-        self.send_response(200)
-        self.send_header("Content-Type", "text/plain; version=0.0.4; charset=utf-8")
-        self.end_headers()
-        self.wfile.write(body.encode("utf-8"))
-
-    def log_message(self, format, *args):
-        logger.debug("%s - - %s", self.address_string(), format % args)
-
-
-def main():
-    port = int(os.environ.get("LLAMA_EXPORTER_PORT", "9550"))
-    host = os.environ.get("LLAMA_EXPORTER_BIND", "0.0.0.0")
-
-    server = HTTPServer((host, port), MetricsHandler)
-    logger.info("Starting Llama Exporter on %s:%d", host, port)
-    logger.info("Polling every %d seconds, timeout %ds", POLL_INTERVAL, SCRAPE_TIMEOUT)
-
-    try:
-        server.serve_forever()
-    except KeyboardInterrupt:
-        logger.info("Shutting down")
-        server.shutdown()
-
-
-if __name__ == "__main__":
-    main()
@@ -18,10 +18,3 @@
    name: "{{ amd_exporter_service_name }}"
    state: restarted
  listen: Restart amd_exporter
-
- name: Restart llama_exporter
-  become: true
-  ansible.builtin.systemd:
-    name: "{{ llama_exporter_service_name }}"
-    state: restarted
-  listen: Restart llama_exporter
@@ -8,4 +8,3 @@
        name: "gpu_monitoring"
      vars:
        amd_exporter_enabled: false
-        llama_exporter_enabled: false
@@ -1,82 +0,0 @@
---
- name: Create llama_exporter group
-  become: true
-  ansible.builtin.group:
-    name: "{{ llama_exporter_user }}"
-    state: present
-    system: true
-
- name: Create llama_exporter user
-  become: true
-  ansible.builtin.user:
-    name: "{{ llama_exporter_user }}"
-    group: "{{ llama_exporter_user }}"
-    home: "{{ llama_exporter_user_home }}"
-    create_home: true
-    shell: /usr/sbin/nologin
-    system: true
-
- name: Create llama_exporter install directory
-  become: true
-  ansible.builtin.file:
-    path: "{{ item }}"
-    state: directory
-    owner: "{{ llama_exporter_user }}"
-    group: "{{ llama_exporter_user }}"
-    mode: "0755"
-  loop:
-    - "{{ llama_exporter_install_dir }}"
-    - "{{ llama_exporter_install_dir }}/logs"
-
- name: Copy llama_exporter script
-  become: true
-  ansible.builtin.copy:
-    src: llama_exporter.py
-    dest: "{{ llama_exporter_script_path }}"
-    owner: "{{ llama_exporter_user }}"
-    group: "{{ llama_exporter_user }}"
-    mode: "0755"
-  notify:
-    - Restart llama_exporter
-
- name: Create Python virtual environment
-  become: true
-  ansible.builtin.command:
-    cmd: "{{ ansible_facts['python']['executable'] }} -m venv {{ llama_exporter_venv_path }}"
-    creates: "{{ llama_exporter_venv_path }}/bin/activate"
-
- name: Install Python dependencies in llama_exporter venv
-  become: true
-  ansible.builtin.pip:
-    name:
-      - prometheus-client
-      - requests
-    virtualenv: "{{ llama_exporter_venv_path }}"
-    state: present
-
- name: Fix ownership of llama_exporter directory
-  become: true
-  ansible.builtin.file:
-    path: "{{ llama_exporter_install_dir }}"
-    owner: "{{ llama_exporter_user }}"
-    group: "{{ llama_exporter_user }}"
-    recurse: true
-
- name: Generate llama_exporter systemd service
-  become: true
-  ansible.builtin.template:
-    src: llama_exporter.service.j2
-    dest: "/etc/systemd/system/{{ llama_exporter_service_name }}.service"
-    owner: root
-    group: root
-    mode: "0644"
-  notify:
-    - Reload systemd daemon
-    - Restart llama_exporter
-
- name: Enable llama_exporter service
-  become: true
-  ansible.builtin.systemd:
-    name: "{{ llama_exporter_service_name }}"
-    enabled: true
-    state: started
@@ -11,12 +11,6 @@
  tags:
    - roles::gpu_monitoring::amd_exporter

- name: Include Llama Exporter setup
-  when: llama_exporter_enabled | default(true) | bool
-  ansible.builtin.include_tasks: llama_exporter.yml
-  tags:
-    - roles::gpu_monitoring::llama_exporter
-
 - name: Include Prometheus setup
  ansible.builtin.include_tasks: prometheus.yml
  tags:
@@ -1,35 +0,0 @@
-[Unit]
-Description=LLM Inference Metrics Exporter (llama.cpp + Ollama)
-After=network-online.target
-Wants=network-online.target
-
-[Service]
-Type=simple
-User={{ llama_exporter_user }}
-Group={{ llama_exporter_user }}
-WorkingDirectory={{ llama_exporter_install_dir }}
-Environment="LLAMA_EXPORTER_PORT={{ llama_exporter_port }}"
-Environment="LLAMA_EXPORTER_BIND=0.0.0.0"
-Environment="LLAMA_EXPORTER_INTERVAL={{ llama_exporter_scrape_interval }}"
-Environment="LLAMA_EXPORTER_TIMEOUT={{ llama_exporter_scrape_timeout }}"
-{% if llama_exporter_targets and llama_exporter_targets != "" %}
-Environment="LLAMA_TARGETS={{ llama_exporter_targets | to_json }}"
-{% endif %}
-Environment="PATH={{ llama_exporter_venv_bin }}:/usr/bin:/bin"
-ExecStart={{ llama_exporter_venv_bin }}/python3 {{ llama_exporter_script_path }}
-Restart=always
-RestartSec=10
-TimeoutStopSec=10
-StandardOutput=journal
-StandardError=journal
-SyslogIdentifier=llama_exporter
-
-# Security hardening
-ProtectSystem=strict
-ProtectHome=true
-ReadWritePaths={{ llama_exporter_install_dir }} /var/log
-NoNewPrivileges=true
-PrivateTmp=true
-
-[Install]
-WantedBy=multi-user.target
@@ -21,10 +21,3 @@ scrape_configs:
      - targets: ["{{ prometheus_host_gateway }}:{{ amd_exporter_port }}"]
        labels:
          datacenter: "{{ inventory_hostname }}"
-
-  - job_name: llama_exporter
-    metrics_path: /metrics
-    static_configs:
-      - targets: ["{{ prometheus_host_gateway }}:{{ llama_exporter_port }}"]
-        labels:
-          datacenter: "{{ inventory_hostname }}"
@@ -12,12 +12,3 @@ nvidia_exporter_service_name: "nvidia_exporter"
 amd_exporter_script_path: "{{ amd_exporter_install_dir }}/amd_gpu_exporter.py"
 amd_exporter_venv_bin: "{{ amd_exporter_venv_path }}/bin"
 amd_exporter_service_name: "amdgpu_exporter"
-
-llama_exporter_script_path: "{{ llama_exporter_install_dir }}/llama_exporter.py"
-llama_exporter_venv_bin: "{{ llama_exporter_venv_path }}/bin"
-llama_exporter_service_name: "llama_exporter"
-
-# Default llama.cpp targets if none provided
-llama_exporter_targets_default:
-  - name: "llama-cpp-8012"
-    url: "http://localhost:8012"