Flow Deployments

Overview

EdgeFlow supports managed flow deployments from the cloud to edge devices. Flows are authored and published in the cloud, then deployed to one or more devices using configurable strategies. Each deployment is tracked with full status reporting, health checks, and automatic rollback on failure.

Flow Lifecycle

┌─────────┐   edit   ┌───────────┐   publish   ┌───────────┐   deploy   ┌──────────┐
│  Draft  │────────>│   Draft   │───────────>│ Published │──────────>│ Deployed │
└─────────┘         └───────────┘            └───────────┘           └──────────┘
                                                   │
                                              archive
                                                   │
                                                   ▼
                                             ┌───────────┐
                                             │ Archived  │
                                             └───────────┘

Flow States

Cloud Flow Management

# Create a flow
POST /api/v1/flows
{
  "name": "Temperature Alert",
  "description": "Monitor temperature and send alerts",
  "tags": ["temperature", "alerts"],
  "definition": {
    "nodes": [
      {"id": "n1", "type": "dht", "name": "DHT22 Sensor", "config": {}, "x": 100, "y": 200},
      {"id": "n2", "type": "switch", "name": "Threshold", "config": {}, "x": 300, "y": 200},
      {"id": "n3", "type": "mqtt_out", "name": "Alert", "config": {}, "x": 500, "y": 200}
    ],
    "connections": [
      {"source": "n1", "target": "n2"},
      {"source": "n2", "target": "n3"}
    ],
    "variables": [
      {"name": "threshold", "type": "number", "value": "30"}
    ]
  }
}

# Publish a flow (makes it deployable)
POST /api/v1/flows/:id/publish

# List all flows
GET /api/v1/flows

# Update a flow (draft only)
PUT /api/v1/flows/:id

# Delete a flow
DELETE /api/v1/flows/:id

Deployment Strategies

All-at-Once

Deploy to all target devices simultaneously. Best for small fleets or non-critical updates.

Rolling

Deploy in sequential batches. If any batch exceeds the failure threshold, deployment pauses and rolls back.

Canary

Deploy to a small percentage of devices first. After a monitoring period, deploy to the rest.

Blue-Green

Deploy to a parallel environment, verify, then switch traffic. Provides instant rollback capability.

Deployment Status

Deployment API

# Create deployment
POST /api/v1/deployments
{
  "flow_id": "flow_abc123",
  "strategy": "rolling",
  "target_devices": ["dev_001", "dev_002", "dev_003"],
  "config": {
    "batch_size": 5,
    "batch_delay": "5m",
    "auto_rollback": true,
    "health_check": true
  }
}

# Start deployment
POST /api/v1/deployments/:id/start

# Check deployment status
GET /api/v1/deployments/:id
{
  "id": "dep_xyz789",
  "flow_id": "flow_abc123",
  "strategy": "rolling",
  "status": "running",
  "total_targets": 3,
  "completed_targets": 1,
  "failed_targets": 0,
  "targets": [
    {"device_id": "dev_001", "status": "success"},
    {"device_id": "dev_002", "status": "deploying"},
    {"device_id": "dev_003", "status": "pending"}
  ],
  "created_at": "2026-02-21T12:00:00Z",
  "started_at": "2026-02-21T12:01:00Z"
}

# Pause deployment
POST /api/v1/deployments/:id/pause

# Rollback deployment
POST /api/v1/deployments/:id/rollback

# List all deployments
GET /api/v1/deployments

Auto-Rollback

When enabled, auto-rollback automatically reverts failed deployments. For rolling deployments, rollback triggers when the failure rate exceeds 20% of any batch. For canary deployments, any failure in the canary group triggers a full rollback.