stemedb/docs/operations/monitoring/grafana/sli-dashboard.json

{
  "dashboard": {
    "title": "StemeDB - SLI & Availability",
    "tags": ["stemedb", "sli", "availability"],
    "timezone": "browser",
    "panels": [
      {
        "id": 1,
        "title": "Request Rate (by endpoint)",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(stemedb_http_requests_total[5m])",
            "legendFormat": "{{method}} {{path}}"
          }
        ],
        "yaxes": [
          {"format": "reqps", "label": "Requests/sec"},
          {"format": "short"}
        ],
        "gridPos": {"h": 8, "w": 12, "x": 0, "y": 0}
      },
      {
        "id": 2,
        "title": "Request Latency p99 (by endpoint)",
        "type": "heatmap",
        "targets": [
          {
            "expr": "histogram_quantile(0.99, rate(stemedb_http_request_duration_seconds_bucket[5m]))",
            "legendFormat": "{{method}} {{path}}"
          }
        ],
        "yaxis": {"format": "s"},
        "gridPos": {"h": 8, "w": 12, "x": 12, "y": 0}
      },
      {
        "id": 3,
        "title": "Error Rate (by type)",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(stemedb_errors_total[5m])",
            "legendFormat": "{{type}} ({{layer}})"
          }
        ],
        "yaxes": [
          {"format": "ops", "label": "Errors/sec"},
          {"format": "short"}
        ],
        "gridPos": {"h": 8, "w": 12, "x": 0, "y": 8},
        "alert": {
          "conditions": [
            {
              "evaluator": {"params": [0.01], "type": "gt"},
              "operator": {"type": "and"},
              "query": {"params": ["A", "5m", "now"]},
              "reducer": {"type": "avg"}
            }
          ],
          "name": "High Error Rate"
        }
      },
      {
        "id": 4,
        "title": "Availability (Success Rate)",
        "type": "gauge",
        "targets": [
          {
            "expr": "sum(rate(stemedb_http_request_duration_seconds_count{status=~\"2..\"}[5m])) / sum(rate(stemedb_http_request_duration_seconds_count[5m]))",
            "legendFormat": "Availability %"
          }
        ],
        "fieldConfig": {
          "defaults": {
            "unit": "percentunit",
            "min": 0,
            "max": 1,
            "thresholds": {
              "mode": "percentage",
              "steps": [
                {"value": 0, "color": "red"},
                {"value": 0.95, "color": "yellow"},
                {"value": 0.99, "color": "green"}
              ]
            }
          }
        },
        "gridPos": {"h": 8, "w": 6, "x": 12, "y": 8}
      },
      {
        "id": 5,
        "title": "Request Status Distribution",
        "type": "piechart",
        "targets": [
          {
            "expr": "sum by (status) (rate(stemedb_http_request_duration_seconds_count[5m]))",
            "legendFormat": "{{status}}"
          }
        ],
        "gridPos": {"h": 8, "w": 6, "x": 18, "y": 8}
      },
      {
        "id": 6,
        "title": "Latency Distribution (all endpoints)",
        "type": "graph",
        "targets": [
          {
            "expr": "histogram_quantile(0.50, rate(stemedb_http_request_duration_seconds_bucket[5m]))",
            "legendFormat": "p50"
          },
          {
            "expr": "histogram_quantile(0.95, rate(stemedb_http_request_duration_seconds_bucket[5m]))",
            "legendFormat": "p95"
          },
          {
            "expr": "histogram_quantile(0.99, rate(stemedb_http_request_duration_seconds_bucket[5m]))",
            "legendFormat": "p99"
          }
        ],
        "yaxes": [
          {"format": "s", "label": "Latency"},
          {"format": "short"}
        ],
        "gridPos": {"h": 8, "w": 12, "x": 0, "y": 16}
      },
      {
        "id": 7,
        "title": "Circuit Breaker Status",
        "type": "stat",
        "targets": [
          {
            "expr": "stemedb_circuit_breakers_open",
            "legendFormat": "Open"
          },
          {
            "expr": "stemedb_circuit_breakers_half_open",
            "legendFormat": "Half-Open"
          }
        ],
        "fieldConfig": {
          "defaults": {
            "unit": "short",
            "thresholds": {
              "mode": "absolute",
              "steps": [
                {"value": 0, "color": "green"},
                {"value": 1, "color": "yellow"},
                {"value": 3, "color": "red"}
              ]
            }
          }
        },
        "gridPos": {"h": 8, "w": 12, "x": 12, "y": 16}
      }
    ],
    "refresh": "15s",
    "schemaVersion": 30,
    "version": 1
  }
}