feat(monitoring): add nginx, postgres, valkey dashboards

roles/grafana/files/nginx-dashboard.json

@@ -0,0 +1,84 @@
+{
+  "title": "Nginx",
+  "uid": "nginx-overview",
+  "tags": ["nginx", "reverse-proxy"],
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "1m",
+  "time": { "from": "now-6h", "to": "now" },
+  "timezone": "browser",
+  "editable": true,
+  "graphTooltip": 1,
+  "templating": {
+    "list": [
+      {
+        "name": "instance",
+        "label": "Host",
+        "type": "query",
+        "datasource": { "type": "prometheus", "uid": "prometheus" },
+        "query": { "query": "label_values(up{job=\"nginx\"}, instance)", "refId": "A" },
+        "refresh": 2,
+        "multi": false,
+        "includeAll": false,
+        "sort": 1,
+        "current": { "selected": false, "text": "", "value": "" }
+      }
+    ]
+  },
+  "panels": [
+    {
+      "id": 1,
+      "type": "stat",
+      "title": "Active Connections",
+      "gridPos": { "h": 4, "w": 6, "x": 0, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "short", "color": { "mode": "thresholds" } }, "overrides": [] },
+      "options": { "reduceOptions": { "values": false, "calcs": ["lastNotNull"], "fields": "" }, "colorMode": "none", "graphMode": "none", "textMode": "auto", "justifyMode": "auto", "orientation": "auto" },
+      "targets": [ { "refId": "A", "expr": "nginx_connections_active{instance=\"$instance\"}" } ]
+    },
+    {
+      "id": 2,
+      "type": "stat",
+      "title": "Requests (6h)",
+      "gridPos": { "h": 4, "w": 6, "x": 6, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "short", "color": { "mode": "thresholds" } }, "overrides": [] },
+      "options": { "reduceOptions": { "values": false, "calcs": ["lastNotNull"], "fields": "" }, "colorMode": "none", "graphMode": "none", "textMode": "auto", "justifyMode": "auto", "orientation": "auto" },
+      "targets": [ { "refId": "A", "expr": "sum(increase(nginx_http_requests_total{instance=\"$instance\"}[$__range]))" } ]
+    },
+    {
+      "id": 3,
+      "type": "timeseries",
+      "title": "Request Rate",
+      "gridPos": { "h": 4, "w": 12, "x": 12, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "reqps", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "sum(rate(nginx_http_requests_total{instance=\"$instance\"}[$__rate_interval]))", "legendFormat": "requests/s" } ]
+    },
+    {
+      "id": 4,
+      "type": "timeseries",
+      "title": "Connection States",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "short", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [
+        { "refId": "A", "expr": "nginx_connections_reading{instance=\"$instance\"}", "legendFormat": "reading" },
+        { "refId": "B", "expr": "nginx_connections_writing{instance=\"$instance\"}", "legendFormat": "writing" },
+        { "refId": "C", "expr": "nginx_connections_waiting{instance=\"$instance\"}", "legendFormat": "waiting" }
+      ]
+    },
+    {
+      "id": 5,
+      "type": "timeseries",
+      "title": "Connections Accepted / Handled",
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "cps", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 10, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [
+        { "refId": "A", "expr": "rate(nginx_connections_accepted{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "accepted" },
+        { "refId": "B", "expr": "rate(nginx_connections_handled{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "handled" }
+      ]
+    }
+  ]
+}

roles/grafana/files/postgres-dashboard.json

@@ -0,0 +1,79 @@
+{
+  "title": "PostgreSQL",
+  "uid": "postgres-overview",
+  "tags": ["postgres", "database"],
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "1m",
+  "time": { "from": "now-6h", "to": "now" },
+  "timezone": "browser",
+  "editable": true,
+  "graphTooltip": 1,
+  "templating": {
+    "list": [
+      {
+        "name": "instance",
+        "label": "Host",
+        "type": "query",
+        "datasource": { "type": "prometheus", "uid": "prometheus" },
+        "query": { "query": "label_values(up{job=\"postgres\"}, instance)", "refId": "A" },
+        "refresh": 2,
+        "multi": false,
+        "includeAll": false,
+        "sort": 1,
+        "current": { "selected": false, "text": "", "value": "" }
+      }
+    ]
+  },
+  "panels": [
+    {
+      "id": 1,
+      "type": "stat",
+      "title": "Up",
+      "gridPos": { "h": 4, "w": 6, "x": 0, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "none", "color": { "mode": "thresholds" }, "thresholds": { "mode": "absolute", "steps": [ { "color": "red", "value": null }, { "color": "green", "value": 1 } ] }, "mappings": [ { "type": "value", "options": { "0": { "text": "Down", "color": "red" }, "1": { "text": "Up", "color": "green" } } } ] }, "overrides": [] },
+      "options": { "reduceOptions": { "values": false, "calcs": ["lastNotNull"], "fields": "" }, "colorMode": "background", "graphMode": "none", "textMode": "auto", "justifyMode": "auto", "orientation": "auto" },
+      "targets": [ { "refId": "A", "expr": "pg_up{instance=\"$instance\"}" } ]
+    },
+    {
+      "id": 2,
+      "type": "timeseries",
+      "title": "Active Connections",
+      "gridPos": { "h": 4, "w": 18, "x": 6, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "short", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "sum by (datname) (pg_stat_database_numbackends{instance=\"$instance\"})", "legendFormat": "{{datname}}" } ]
+    },
+    {
+      "id": 3,
+      "type": "timeseries",
+      "title": "Cache Hit Ratio",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "percentunit", "min": 0, "max": 1, "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "sum(rate(pg_stat_database_blks_hit{instance=\"$instance\"}[$__rate_interval])) / (sum(rate(pg_stat_database_blks_hit{instance=\"$instance\"}[$__rate_interval])) + sum(rate(pg_stat_database_blks_read{instance=\"$instance\"}[$__rate_interval])))", "legendFormat": "hit ratio" } ]
+    },
+    {
+      "id": 4,
+      "type": "timeseries",
+      "title": "Transactions",
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "ops", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 10, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [
+        { "refId": "A", "expr": "rate(pg_stat_database_xact_commit{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "commit" },
+        { "refId": "B", "expr": "rate(pg_stat_database_xact_rollback{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "rollback" }
+      ]
+    },
+    {
+      "id": 5,
+      "type": "timeseries",
+      "title": "Deadlocks",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 12 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "ops", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "rate(pg_stat_database_deadlocks{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "{{datname}}" } ]
+    }
+  ]
+}

roles/grafana/files/valkey-dashboard.json

@@ -0,0 +1,88 @@
+{
+  "title": "Valkey",
+  "uid": "valkey-overview",
+  "tags": ["valkey", "redis", "cache"],
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "1m",
+  "time": { "from": "now-6h", "to": "now" },
+  "timezone": "browser",
+  "editable": true,
+  "graphTooltip": 1,
+  "templating": {
+    "list": [
+      {
+        "name": "instance",
+        "label": "Host",
+        "type": "query",
+        "datasource": { "type": "prometheus", "uid": "prometheus" },
+        "query": { "query": "label_values(up{job=\"valkey\"}, instance)", "refId": "A" },
+        "refresh": 2,
+        "multi": false,
+        "includeAll": false,
+        "sort": 1,
+        "current": { "selected": false, "text": "", "value": "" }
+      }
+    ]
+  },
+  "panels": [
+    {
+      "id": 1,
+      "type": "stat",
+      "title": "Up",
+      "gridPos": { "h": 4, "w": 6, "x": 0, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "none", "color": { "mode": "thresholds" }, "thresholds": { "mode": "absolute", "steps": [ { "color": "red", "value": null }, { "color": "green", "value": 1 } ] }, "mappings": [ { "type": "value", "options": { "0": { "text": "Down", "color": "red" }, "1": { "text": "Up", "color": "green" } } } ] }, "overrides": [] },
+      "options": { "reduceOptions": { "values": false, "calcs": ["lastNotNull"], "fields": "" }, "colorMode": "background", "graphMode": "none", "textMode": "auto", "justifyMode": "auto", "orientation": "auto" },
+      "targets": [ { "refId": "A", "expr": "redis_up{instance=\"$instance\"}" } ]
+    },
+    {
+      "id": 2,
+      "type": "timeseries",
+      "title": "Connected Clients",
+      "gridPos": { "h": 4, "w": 18, "x": 6, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "short", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "redis_connected_clients{instance=\"$instance\"}", "legendFormat": "clients" } ]
+    },
+    {
+      "id": 3,
+      "type": "timeseries",
+      "title": "Memory",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "bytes", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [
+        { "refId": "A", "expr": "redis_memory_used_bytes{instance=\"$instance\"}", "legendFormat": "used" },
+        { "refId": "B", "expr": "redis_memory_max_bytes{instance=\"$instance\"}", "legendFormat": "max" }
+      ]
+    },
+    {
+      "id": 4,
+      "type": "timeseries",
+      "title": "Ops/sec",
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 4 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "ops", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "rate(redis_commands_processed_total{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "ops/s" } ]
+    },
+    {
+      "id": 5,
+      "type": "timeseries",
+      "title": "Cache Hit Rate",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 12 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "percentunit", "min": 0, "max": 1, "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "rate(redis_keyspace_hits_total{instance=\"$instance\"}[$__rate_interval]) / (rate(redis_keyspace_hits_total{instance=\"$instance\"}[$__rate_interval]) + rate(redis_keyspace_misses_total{instance=\"$instance\"}[$__rate_interval]))", "legendFormat": "hit rate" } ]
+    },
+    {
+      "id": 6,
+      "type": "timeseries",
+      "title": "Evicted Keys",
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 12 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "ops", "color": { "mode": "palette-classic" }, "custom": { "fillOpacity": 20, "lineWidth": 1, "drawStyle": "line", "showPoints": "never" } }, "overrides": [] },
+      "targets": [ { "refId": "A", "expr": "rate(redis_evicted_keys_total{instance=\"$instance\"}[$__rate_interval])", "legendFormat": "evicted/s" } ]
+    }
+  ]
+}

roles/grafana/tasks/main.yml

@@ -209,6 +209,25 @@
     mode: "0644"
   notify: Restart grafana
 
+- name: Remove legacy combined services dashboard
+  ansible.builtin.file:
+    path: "{{ grafana_provisioning_path }}/dashboards/services.json"
+    state: absent
+  notify: Restart grafana
+
+- name: Deploy services dashboards (nginx, postgres, valkey)
+  ansible.builtin.copy:
+    src: "{{ item }}"
+    dest: "{{ grafana_provisioning_path }}/dashboards/{{ item }}"
+    owner: "{{ grafana_user }}"
+    group: "{{ grafana_group }}"
+    mode: "0644"
+  loop:
+    - nginx-dashboard.json
+    - postgres-dashboard.json
+    - valkey-dashboard.json
+  notify: Restart grafana
+
 # Service management
 - name: Enable and start Grafana service
   ansible.builtin.systemd: