jesus/Vapora
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity
Vapora/provisioning/vapora-wrksp/workflows/deploy-full-stack.yaml

329 lines
12 KiB
YAML
Raw Normal View History

feat: Phase 5.3 - Multi-Agent Learning Infrastructure Implement intelligent agent learning from Knowledge Graph execution history with per-task-type expertise tracking, recency bias, and learning curves. ## Phase 5.3 Implementation ### Learning Infrastructure (✅ Complete) - LearningProfileService with per-task-type expertise metrics - TaskTypeExpertise model tracking success_rate, confidence, learning curves - Recency bias weighting: recent 7 days weighted 3x higher (exponential decay) - Confidence scoring prevents overfitting: min(1.0, executions / 20) - Learning curves computed from daily execution windows ### Agent Scoring Service (✅ Complete) - Unified AgentScore combining SwarmCoordinator + learning profiles - Scoring formula: 0.3*base + 0.5*expertise + 0.2*confidence - Rank agents by combined score for intelligent assignment - Support for recency-biased scoring (recent_success_rate) - Methods: rank_agents, select_best, rank_agents_with_recency ### KG Integration (✅ Complete) - KGPersistence::get_executions_for_task_type() - query by agent + task type - KGPersistence::get_agent_executions() - all executions for agent - Coordinator::load_learning_profile_from_kg() - core KG→Learning integration - Coordinator::load_all_learning_profiles() - batch load for multiple agents - Convert PersistedExecution → ExecutionData for learning calculations ### Agent Assignment Integration (✅ Complete) - AgentCoordinator uses learning profiles for task assignment - extract_task_type() infers task type from title/description - assign_task() scores candidates using AgentScoringService - Fallback to load-based selection if no learning data available - Learning profiles stored in coordinator.learning_profiles RwLock ### Profile Adapter Enhancements (✅ Complete) - create_learning_profile() - initialize empty profiles - add_task_type_expertise() - set task-type expertise - update_profile_with_learning() - update swarm profiles from learning ## Files Modified ### vapora-knowledge-graph/src/persistence.rs (+30 lines) - get_executions_for_task_type(agent_id, task_type, limit) - get_agent_executions(agent_id, limit) ### vapora-agents/src/coordinator.rs (+100 lines) - load_learning_profile_from_kg() - core KG integration method - load_all_learning_profiles() - batch loading for agents - assign_task() already uses learning-based scoring via AgentScoringService ### Existing Complete Implementation - vapora-knowledge-graph/src/learning.rs - calculation functions - vapora-agents/src/learning_profile.rs - data structures and expertise - vapora-agents/src/scoring.rs - unified scoring service - vapora-agents/src/profile_adapter.rs - adapter methods ## Tests Passing - learning_profile: 7 tests ✅ - scoring: 5 tests ✅ - profile_adapter: 6 tests ✅ - coordinator: learning-specific tests ✅ ## Data Flow 1. Task arrives → AgentCoordinator::assign_task() 2. Extract task_type from description 3. Query KG for task-type executions (load_learning_profile_from_kg) 4. Calculate expertise with recency bias 5. Score candidates (SwarmCoordinator + learning) 6. Assign to top-scored agent 7. Execution result → KG → Update learning profiles ## Key Design Decisions ✅ Recency bias: 7-day half-life with 3x weight for recent performance ✅ Confidence scoring: min(1.0, total_executions / 20) prevents overfitting ✅ Hierarchical scoring: 30% base load, 50% expertise, 20% confidence ✅ KG query limit: 100 recent executions per task-type for performance ✅ Async loading: load_learning_profile_from_kg supports concurrent loads ## Next: Phase 5.4 - Cost Optimization Ready to implement budget enforcement and cost-aware provider selection.
2026-01-11 13:03:53 +00:00
apiVersion: provisioning.vapora.io/v1
kind: Workflow
metadata:
name: deploy-full-stack
description: Complete VAPORA deployment from scratch including cluster, databases, and services
spec:
# Workflow metadata
version: "0.2.0"
namespace: vapora-system
timeout: 3600s # 1 hour max
retryPolicy:
maxRetries: 3
backoffFactor: 2
# Prerequisites
prerequisites:
- kubeconfig_present
- provisioning_cli_installed
- sufficient_resources:
cpu: "20"
memory: "64Gi"
disk: "500Gi"
# Workflow phases executed sequentially with gates
phases:
# Phase 1: Infrastructure foundation
- name: "Create K8s Cluster"
description: "Deploy base Kubernetes cluster with networking"
retryable: true
steps:
- name: "Apply KCL cluster schema"
command: "provisioning cluster create --config kcl/cluster.k"
timeout: 1200s
onError: "rollback_cluster"
- name: "Install CNI (Cilium)"
command: "provisioning addon install cilium --helm-values cilium-values.yaml"
timeout: 300s
retries: 3
- name: "Install service mesh (Istio)"
command: "provisioning addon install istio --config kcl/cluster.k"
timeout: 600s
dependencies: ["cilium"]
- name: "Install storage (Rook Ceph)"
command: "provisioning addon install rook-ceph --size 500Gi --replicas 3"
timeout: 900s
dependencies: ["cilium"]
- name: "Verify cluster health"
command: "provisioning health-check --cluster"
timeout: 300s
dependencies: ["cilium", "istio", "rook-ceph"]
# Phase 2: Create namespaces and RBAC
- name: "Setup Namespaces and Security"
description: "Create namespaces, service accounts, and RBAC policies"
retryable: true
steps:
- name: "Create namespaces"
command: "kubectl apply -f - <<EOF\napiVersion: v1\nkind: Namespace\nmetadata:\n name: vapora-system\n labels:\n app: vapora\nEOF"
timeout: 60s
dependencies: ["Create K8s Cluster"]
- name: "Create service accounts"
command: "kubectl apply -f - <<EOF\napiVersion: v1\nkind: ServiceAccount\nmetadata:\n name: vapora-backend\n namespace: vapora-system\n---\napiVersion: v1\nkind: ServiceAccount\nmetadata:\n name: vapora-agents\n namespace: vapora-system\nEOF"
timeout: 60s
dependencies: ["Create namespaces"]
- name: "Apply RBAC policies"
command: "provisioning rbac apply --config kcl/services.k --namespace vapora-system"
timeout: 120s
dependencies: ["Create service accounts"]
# Phase 3: Deploy storage layer
- name: "Deploy Storage Infrastructure"
description: "Deploy SurrealDB, Redis, and NATS"
retryable: true
parallel: false # Must be sequential for data consistency
steps:
- name: "Deploy SurrealDB"
command: "provisioning taskserv deploy --config taskservs/vapora-storage.toml --component surrealdb"
timeout: 600s
dependencies: ["Setup Namespaces and Security"]
healthCheck:
type: "http"
endpoint: "http://surrealdb-0.surrealdb.vapora-system:8000/health"
interval: 10s
maxRetries: 30
- name: "Deploy Redis"
command: "provisioning taskserv deploy --config taskservs/vapora-storage.toml --component redis"
timeout: 300s
dependencies: ["Deploy SurrealDB"]
healthCheck:
type: "exec"
command: "redis-cli ping"
interval: 5s
maxRetries: 20
- name: "Deploy NATS JetStream"
command: "provisioning taskserv deploy --config taskservs/vapora-storage.toml --component nats"
timeout: 300s
dependencies: ["Deploy Redis"]
healthCheck:
type: "http"
endpoint: "http://nats-0:8222/varz"
interval: 10s
maxRetries: 20
- name: "Initialize database schema"
command: "provisioning db init --database surrealdb --schema-dir scripts/migrations"
timeout: 180s
dependencies: ["Deploy SurrealDB"]
# Phase 4: Deploy backend services
- name: "Deploy Backend Services"
description: "Deploy Axum backend, LLM router, and MCP gateway"
retryable: true
parallel: true # Can deploy in parallel
steps:
- name: "Deploy REST API backend"
command: "provisioning taskserv deploy vapora-backend"
timeout: 300s
dependencies: ["Deploy Storage Infrastructure"]
healthCheck:
type: "http"
endpoint: "http://vapora-backend.vapora-system:8080/api/v1/health"
interval: 10s
maxRetries: 20
- name: "Deploy Multi-IA LLM Router"
command: "provisioning taskserv deploy vapora-llm-router"
timeout: 300s
dependencies: ["Deploy Storage Infrastructure"]
healthCheck:
type: "http"
endpoint: "http://vapora-llm-router.vapora-system:8899/health"
interval: 10s
maxRetries: 20
- name: "Deploy MCP Gateway"
command: "provisioning taskserv deploy vapora-mcp-gateway"
timeout: 300s
dependencies: ["Deploy Storage Infrastructure"]
healthCheck:
type: "http"
endpoint: "http://vapora-mcp-gateway.vapora-system:8888/health"
interval: 10s
maxRetries: 20
# Phase 5: Deploy agent runtime
- name: "Deploy Agent Runtime"
description: "Deploy 12-agent orchestrator with initial replicas"
retryable: true
steps:
- name: "Deploy agent runtime pods"
command: "provisioning taskserv deploy vapora-agents --replicas 3"
timeout: 600s
dependencies: ["Deploy Backend Services"]
healthCheck:
type: "custom"
script: |
AGENT_COUNT=$(kubectl get pods -n vapora-system -l app=vapora-agents --field-selector=status.phase=Running | wc -l)
if [ "$AGENT_COUNT" -ge 2 ]; then
echo "OK: $AGENT_COUNT agents running"
exit 0
else
echo "ERROR: Only $AGENT_COUNT agents running"
exit 1
fi
interval: 30s
maxRetries: 20
# Phase 6: Deploy frontend
- name: "Deploy Frontend"
description: "Deploy Leptos CSR frontend"
retryable: true
steps:
- name: "Deploy frontend application"
command: "provisioning taskserv deploy vapora-frontend"
timeout: 300s
dependencies: ["Deploy Agent Runtime"]
healthCheck:
type: "http"
endpoint: "http://vapora-frontend.vapora-system:3000/"
interval: 10s
maxRetries: 20
# Phase 7: Setup monitoring and observability
- name: "Setup Monitoring Stack"
description: "Deploy Prometheus, Grafana, and Loki"
retryable: true
parallel: true
steps:
- name: "Deploy Prometheus"
command: "provisioning addon install prometheus --namespace monitoring"
timeout: 300s
dependencies: ["Deploy Frontend"]
- name: "Deploy Grafana"
command: "provisioning addon install grafana --namespace monitoring"
timeout: 300s
dependencies: ["Deploy Frontend"]
- name: "Deploy Loki (log aggregation)"
command: "provisioning addon install loki --namespace monitoring"
timeout: 300s
dependencies: ["Deploy Frontend"]
# Phase 8: Configure networking and ingress
- name: "Configure Ingress and Networking"
description: "Setup Istio gateway and ingress rules"
retryable: true
steps:
- name: "Apply Istio gateway configuration"
command: "kubectl apply -f - <<EOF\napiVersion: networking.istio.io/v1beta1\nkind: Gateway\nmetadata:\n name: vapora-gateway\n namespace: vapora-system\nspec:\n selector:\n istio: ingressgateway\n servers:\n - port:\n number: 80\n name: http\n protocol: HTTP\n hosts:\n - \"vapora.example.com\"\n - port:\n number: 443\n name: https\n protocol: HTTPS\n tls:\n mode: SIMPLE\n credentialName: vapora-tls\n hosts:\n - \"vapora.example.com\"\nEOF"
timeout: 60s
dependencies: ["Deploy Frontend"]
- name: "Apply VirtualService routing"
command: "provisioning istio apply --config kcl/cluster.k --namespace vapora-system"
timeout: 120s
dependencies: ["Apply Istio gateway configuration"]
# Phase 9: Post-deployment verification
- name: "Verify Full Stack"
description: "Run comprehensive health checks and smoke tests"
retryable: false
steps:
- name: "Check all services are running"
command: |
provisioning health-check --services all
if [ $? -ne 0 ]; then
echo "ERROR: Some services are not healthy"
exit 1
fi
timeout: 300s
dependencies: ["Configure Ingress and Networking"]
- name: "Run smoke tests"
command: |
provisioning test smoke --api http://vapora-backend.vapora-system:8080
provisioning test smoke --frontend http://vapora-frontend.vapora-system:3000
timeout: 180s
dependencies: ["Check all services are running"]
- name: "Verify database connectivity"
command: "provisioning db test-connection --database surrealdb"
timeout: 60s
dependencies: ["Check all services are running"]
- name: "Verify agent communication"
command: "provisioning agents health-check --nats nats://nats-0.vapora-system:4222"
timeout: 120s
dependencies: ["Check all services are running"]
# Output configuration
outputs:
- name: frontend_url
value: "https://vapora.example.com"
- name: grafana_url
value: "https://vapora.example.com/grafana"
- name: prometheus_url
value: "https://vapora.example.com:9090"
- name: cluster_info
value: "kubectl cluster-info"
# Rollback policies
rollback:
onFailure: "manual" # manual | automatic | rollback-to-previous
allowedSteps: ["Create K8s Cluster", "Deploy Storage Infrastructure", "Deploy Backend Services"]
strategy: "cascade" # cascade | parallel
# Notifications
notifications:
onStart:
- "email: devops@example.com"
- "slack: #deployment"
onSuccess:
- "email: devops@example.com"
- "slack: #deployment"
- "action: update-dns"
onFailure:
- "email: devops@example.com"
- "slack: #deployment"
- "severity: critical"
# Post-deployment actions
postDeployment:
- name: "Create backup"
command: "provisioning backup create --cluster vapora-cluster"
schedule: "daily"
- name: "Generate deployment report"
command: "provisioning report generate --format markdown --output deployment-report-$(date +%s).md"
schedule: "once"
---
# Rollback workflow (automatically created)
apiVersion: provisioning.vapora.io/v1
kind: Workflow
metadata:
name: deploy-full-stack-rollback
description: Rollback VAPORA deployment to previous stable state
spec:
version: "0.2.0"
phases:
- name: "Stop new operations"
steps:
- name: "Drain agent queue"
command: "provisioning agents drain --wait-timeout 300s"
- name: "Restore from backup"
steps:
- name: "List available backups"
command: "provisioning backup list"
- name: "Restore cluster state"
command: "provisioning backup restore --backup latest --confirm"
- name: "Verify rollback"
steps:
- name: "Run health checks"
command: "provisioning health-check --cluster"
Reference in New Issue Copy Permalink