claude-flow-novice/monitoring/alerts/latency-alerts.yml

Claude Flow Novice - Advanced orchestration platform for multi-agent AI workflows with CFN Loop architecture Includes Local RuVector Accelerator and all CFN skills for complete functionality.

groups:
  - name: latency_alerts
    interval: 30s
    rules:
      - alert: HighP99Latency
        expr: |
          histogram_quantile(0.99, rate(http_request_duration_seconds_bucket[5m])) * 1000 > 7500
        for: 10m
        labels:
          severity: critical
          component: integration
        annotations:
          summary: "P99 latency exceeds threshold"
          description: "P99 latency is {{ $value | humanize }}ms (threshold: 7500ms)"
          runbook_url: "docs/INCIDENT_RESPONSE.md"

      - alert: HighP95Latency
        expr: |
          histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m])) * 1000 > 3000
        for: 10m
        labels:
          severity: high
          component: integration
        annotations:
          summary: "P95 latency exceeds threshold"
          description: "P95 latency is {{ $value | humanize }}ms (threshold: 3000ms)"

      - alert: HighP50Latency
        expr: |
          histogram_quantile(0.5, rate(http_request_duration_seconds_bucket[5m])) * 1000 > 750
        for: 15m
        labels:
          severity: warning
          component: integration
        annotations:
          summary: "P50 latency exceeds threshold"
          description: "P50 latency is {{ $value | humanize }}ms (threshold: 750ms)"

      - alert: DatabaseQueryLatency
        expr: |
          histogram_quantile(0.95, rate(db_query_duration_seconds_bucket[5m])) * 1000 > 5000
        for: 10m
        labels:
          severity: high
          component: database
        annotations:
          summary: "Database query latency too high"
          description: "P95 query latency is {{ $value | humanize }}ms (threshold: 5000ms)"
          runbook_url: "docs/ROLLBACK_RUNBOOK.md"

      - alert: DatabaseTransactionLatency
        expr: |
          histogram_quantile(0.99, rate(db_transaction_duration_seconds_bucket[5m])) * 1000 > 10000
        for: 10m
        labels:
          severity: high
          component: database
        annotations:
          summary: "Database transaction latency too high"
          description: "P99 transaction latency is {{ $value | humanize }}ms"

      - alert: CoordinationProtocolLatency
        expr: |
          histogram_quantile(0.95, rate(coordination_protocol_latency_seconds_bucket[5m])) * 1000 > 500
        for: 10m
        labels:
          severity: warning
          component: coordination
        annotations:
          summary: "Coordination protocol latency elevated"
          description: "P95 protocol latency is {{ $value | humanize }}ms (threshold: 500ms)"

      - alert: SkillExecutionLatency
        expr: |
          histogram_quantile(0.95, rate(skill_execution_duration_seconds_bucket[5m])) * 1000 > 5000
        for: 15m
        labels:
          severity: warning
          component: skill_deployment
        annotations:
          summary: "Skill execution latency elevated"
          description: "P95 execution time is {{ $value | humanize }}ms"

      - alert: ArtifactStorageLatency
        expr: |
          histogram_quantile(0.95, rate(artifact_storage_latency_seconds_bucket[5m])) * 1000 > 2000
        for: 10m
        labels:
          severity: warning
          component: artifact_storage
        annotations:
          summary: "Artifact storage latency elevated"
          description: "P95 storage operation latency is {{ $value | humanize }}ms"

      - alert: LatencyIncreaseTwofold
        expr: |
          (histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m])) / on() histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[1h] offset 24h))) > 2
        for: 10m
        labels:
          severity: critical
          component: integration
        annotations:
          summary: "Latency doubled compared to baseline"
          description: "Current latency is 2x baseline - potential degradation"
          runbook_url: "docs/INCIDENT_RESPONSE.md"

      - alert: QueueDepthBuildup
        expr: |
          redis_queue_size > 1000
        for: 10m
        labels:
          severity: high
          component: coordination
        annotations:
          summary: "Queue depth exceeds threshold"
          description: "Current queue depth: {{ $value }} messages (threshold: 1000)"
          impact: "Processing latency may increase"

      - alert: ConnectionPoolWaitTime
        expr: |
          histogram_quantile(0.95, rate(db_connection_wait_seconds_bucket[5m])) * 1000 > 500
        for: 10m
        labels:
          severity: warning
          component: database
        annotations:
          summary: "Database connection pool wait time high"
          description: "P95 wait time is {{ $value | humanize }}ms"