update

Author: tom-sapletta-com

containers/llm-orchestrator-min/ansible/tasks/setup_monitoring.yml

@@ -0,0 +1,60 @@
+---
+# Zadania do konfiguracji monitoringu dla mikrousług llm-orchestrator
+- name: Utworzenie katalogu dla monitoringu
+  file:
+    path: "{{ project_root }}/monitoring"
+    state: directory
+    mode: '0755'
+
+- name: Kopiowanie konfiguracji Prometheus
+  template:
+    src: templates/prometheus.yml.j2
+    dest: "{{ project_root }}/monitoring/prometheus.yml"
+
+- name: Kopiowanie konfiguracji Grafana
+  template:
+    src: templates/grafana.ini.j2
+    dest: "{{ project_root }}/monitoring/grafana.ini"
+
+- name: Kopiowanie konfiguracji dashboardów Grafana
+  copy:
+    src: files/dashboards/
+    dest: "{{ project_root }}/monitoring/dashboards/"
+    directory_mode: yes
+
+- name: Tworzenie docker-compose dla monitoringu
+  template:
+    src: templates/docker-compose-monitoring.yml.j2
+    dest: "{{ project_root }}/monitoring/docker-compose.yml"
+
+- name: Uruchomienie stosu monitoringu
+  community.docker.docker_compose:
+    project_src: "{{ project_root }}/monitoring"
+    state: present
+
+- name: Czekanie na uruchomienie Grafana
+  uri:
+    url: http://localhost:3000/api/health
+    status_code: 200
+  register: result
+  until: result.status == 200
+  retries: 30
+  delay: 5
+  ignore_errors: yes
+
+- name: Konfiguracja źródła danych Prometheus w Grafana
+  uri:
+    url: http://localhost:3000/api/datasources
+    method: POST
+    body_format: json
+    body:
+      name: "Prometheus"
+      type: "prometheus"
+      url: "http://prometheus:9090"
+      access: "proxy"
+      isDefault: true
+    user: "admin"
+    password: "{{ grafana_admin_password }}"
+    force_basic_auth: yes
+    status_code: 200
+  ignore_errors: yes

containers/llm-orchestrator-min/ansible/templates/docker-compose-monitoring.yml.j2

@@ -0,0 +1,78 @@
+version: '3.8'
+
+services:
+  prometheus:
+    image: prom/prometheus:v2.40.0
+    container_name: prometheus
+    volumes:
+      - ./prometheus.yml:/etc/prometheus/prometheus.yml
+      - prometheus_data:/prometheus
+    command:
+      - '--config.file=/etc/prometheus/prometheus.yml'
+      - '--storage.tsdb.path=/prometheus'
+      - '--web.console.libraries=/etc/prometheus/console_libraries'
+      - '--web.console.templates=/etc/prometheus/consoles'
+      - '--web.enable-lifecycle'
+    ports:
+      - "{{ prometheus_port | default(9090) }}:9090"
+    restart: unless-stopped
+    networks:
+      - monitoring-network
+      - llm-network
+
+  grafana:
+    image: grafana/grafana:9.3.0
+    container_name: grafana
+    volumes:
+      - ./grafana.ini:/etc/grafana/grafana.ini
+      - ./dashboards:/var/lib/grafana/dashboards
+      - grafana_data:/var/lib/grafana
+    environment:
+      - GF_SECURITY_ADMIN_PASSWORD={{ grafana_admin_password | default('admin') }}
+      - GF_USERS_ALLOW_SIGN_UP=false
+    ports:
+      - "{{ grafana_port | default(3000) }}:3000"
+    restart: unless-stopped
+    networks:
+      - monitoring-network
+
+  node-exporter:
+    image: prom/node-exporter:v1.4.0
+    container_name: node-exporter
+    volumes:
+      - /proc:/host/proc:ro
+      - /sys:/host/sys:ro
+      - /:/rootfs:ro
+    command:
+      - '--path.procfs=/host/proc'
+      - '--path.rootfs=/rootfs'
+      - '--path.sysfs=/host/sys'
+      - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
+    restart: unless-stopped
+    networks:
+      - monitoring-network
+
+  cadvisor:
+    image: gcr.io/cadvisor/cadvisor:v0.45.0
+    container_name: cadvisor
+    volumes:
+      - /:/rootfs:ro
+      - /var/run:/var/run:ro
+      - /sys:/sys:ro
+      - /var/lib/docker/:/var/lib/docker:ro
+      - /dev/disk/:/dev/disk:ro
+    ports:
+      - "8081:8080"
+    restart: unless-stopped
+    networks:
+      - monitoring-network
+
+networks:
+  monitoring-network:
+    driver: bridge
+  llm-network:
+    external: true
+
+volumes:
+  prometheus_data:
+  grafana_data:

containers/llm-orchestrator-min/ansible/templates/env.j2

@@ -0,0 +1,19 @@
+# Zmienne środowiskowe dla llm-orchestrator-min
+# Generowane automatycznie przez Ansible
+
+# Ogólne
+ENVIRONMENT={{ environment }}
+
+# Model Service
+MODEL_PATH={{ model_path | default('/app/models/tinyllama') }}
+USE_INT8={{ use_int8 | default(true) | string | lower }}
+MODEL_SERVICE_PORT={{ model_service_port | default(5000) }}
+
+# API Gateway
+API_GATEWAY_PORT={{ api_gateway_port | default(80) }}
+DASHBOARD_PORT={{ dashboard_port | default(8080) }}
+
+# Monitoring
+PROMETHEUS_PORT={{ prometheus_port | default(9090) }}
+GRAFANA_PORT={{ grafana_port | default(3000) }}
+GRAFANA_ADMIN_PASSWORD={{ grafana_admin_password | default('admin') }}

containers/llm-orchestrator-min/ansible/templates/prometheus.yml.j2

@@ -0,0 +1,36 @@
+global:
+  scrape_interval: 15s
+  evaluation_interval: 15s
+
+alerting:
+  alertmanagers:
+    - static_configs:
+        - targets:
+          # - alertmanager:9093
+
+rule_files:
+  # - "first_rules.yml"
+  # - "second_rules.yml"
+
+scrape_configs:
+  - job_name: "prometheus"
+    static_configs:
+      - targets: ["localhost:9090"]
+
+  - job_name: "model-service"
+    metrics_path: /metrics
+    static_configs:
+      - targets: ["model-service:5000"]
+
+  - job_name: "api-gateway"
+    metrics_path: /metrics
+    static_configs:
+      - targets: ["api-gateway:8080"]
+
+  - job_name: "node-exporter"
+    static_configs:
+      - targets: ["node-exporter:9100"]
+
+  - job_name: "cadvisor"
+    static_configs:
+      - targets: ["cadvisor:8080"]