Bbobd988doc

f31e7506创建于 2025年11月6日历史提交

Python Distributed Dora Sensor Network

This example demonstrates Dora's built-in distributed deployment capabilities using the coordinator-daemon architecture with automatic Zenoh routing for cross-network communication.

Key Features

✅ Native Dora Distribution: Uses _unstable_deploy.machine tags ✅ Coordinator-Daemon Architecture: Standard Dora distributed pattern ✅ Automatic Zenoh Routing: Cross-network communication handled by Dora ✅ No Manual Zenoh Setup: Dora manages Zenoh internally ✅ Machine-Based Deployment: Each node assigned to specific machine ID

Architecture

                   Dora Coordinator
                          │
         ┌────────────────┼────────────────┐
         │                │                │
    Daemon (edge1)   Daemon (edge2)   Daemon (edge3)   Daemon (cloud)
         │                │                │                 │
    sensor-temp     sensor-humidity  sensor-pressure   cloud-aggregator
    (temperature)     (humidity)      (pressure)       (aggregates all)

Zenoh Integration: When nodes are on different machines, Dora automatically uses its built-in Zenoh backend to route messages. Data flows through Zenoh topics like:

dora/default/{dataflow_id}/output/sensor-temp/data
dora/default/{dataflow_id}/output/sensor-humidity/data
dora/default/{dataflow_id}/output/sensor-pressure/data

Dataflow Configuration

The dataflow.yml uses Dora's _unstable_deploy feature:

nodes:
  - id: sensor-temp
    _unstable_deploy:
      machine: edge1      # Runs on daemon with --machine-id edge1
      working_dir: /Users/username/python-distributed-zenoh
    path: python3
    args: sensor_node.py sensor_temp temperature
    outputs:
      - data

  - id: cloud-aggregator
    _unstable_deploy:
      machine: cloud      # Runs on daemon with --machine-id cloud
      working_dir: /Users/username/python-distributed-zenoh
    inputs:
      temp_data: sensor-temp/data
      humidity_data: sensor-humidity/data
      pressure_data: sensor-pressure/data

Important: The working_dir parameter tells each daemon where to find the Python scripts. For real distributed deployments, ensure the scripts exist at this path on each target machine.

Installation

On All Machines

# Install dora-rs (includes CLI and Python API)
pip install dora-rs

# Optional: Install Zenoh tools for cross-network deployments
# (requires Rust/Cargo)
cargo install zenoh --features unstable

AWS EC2 Security Group Configuration (For Cross-Network)

If deploying to EC2, configure security group to allow:

TCP 7447 - Zenoh router (from 0.0.0.0/0)
TCP 53290 - Dora coordinator (from 0.0.0.0/0)

In AWS Console: EC2 → Security Groups → Add Inbound Rules

Running on ONE Machine (Simulation)

Simulate distributed deployment with multiple daemons on one PC:

Manual Steps

Terminal 1 - Coordinator:

dora coordinator

Terminal 2 - Edge Device 1 (Temperature Sensor):

dora daemon --machine-id edge1

Terminal 3 - Edge Device 2 (Humidity Sensor):

dora daemon --machine-id edge2 --local-listen-port 53291

Terminal 4 - Edge Device 3 (Pressure Sensor):

dora daemon --machine-id edge3 --local-listen-port 53292

Terminal 5 - Cloud Server:

dora daemon --machine-id cloud --local-listen-port 53293

Terminal 6 - Start Dataflow:

dora build dataflow.yml
dora start dataflow.yml

Running on real distribution

Same Network (WiFi/LAN)

All machines in the same network can auto-discover via Zenoh multicast.

On Cloud Server (e.g., 192.168.1.100):

# Start coordinator
dora coordinator

On Each Edge Device:

# Edge device 1
dora daemon --coordinator-addr 192.168.1.100 --machine-id edge1

# Edge device 2
dora daemon --coordinator-addr 192.168.1.100 --machine-id edge2

# Edge device 3
dora daemon --coordinator-addr 192.168.1.100 --machine-id edge3

On Cloud Server:

# This daemon runs the cloud-aggregator node
dora daemon --machine-id cloud

# Start the dataflow
dora start dataflow.yml

Different Networks (Cross-Network) - AWS EC2 Example

When machines are in different networks (behind NAT/firewalls), use Zenoh routers. This example shows connecting a home PC to AWS EC2.

Prerequisites

On Both Machines:

pip install dora-rs

Copy Files to EC2 (~/dora-distributed/):
- sensor_node.py
- cloud_node.py
- dataflow_distributed.yml
- zenoh_config.json5
Update dataflow_distributed.yml paths:
- Edge device: Set working_dir to your local path (e.g., /Users/username/python-distributed-zenoh for Mac, /home/username/dora-distributed for Linux)
- Cloud: Set working_dir to EC2 path (e.g., /home/ubuntu/dora-distributed)

Example configuration:

nodes:
  - id: sensor-temp
    _unstable_deploy:
      machine: edge1
      working_dir: /Users/username/python-distributed-zenoh  # Mac path
    path: python3
    args: sensor_node.py sensor_temp temperature
    inputs:
      tick: dora/timer/millis/1000
    outputs:
      - data

  - id: cloud-aggregator
    _unstable_deploy:
      machine: cloud
      working_dir: /home/ubuntu/dora-distributed  # EC2 path
    path: python3
    args: cloud_node.py
    inputs:
      temp_data: sensor-temp/data

Setup Zenoh Config

Create zenoh_config.json5 (on both machines):

{
  mode: "client",
  connect: {
    endpoints: ["tcp/YOUR_EC2_IP:7447"]  // Replace with your EC2 public IP
  },
  listen: {
    endpoints: []
  }
}

Key Settings:

mode: "client" - Makes dora daemon a Zenoh client (not a router)
listen: { endpoints: [] } - Prevents port conflicts with zenohd router
Replace YOUR_EC2_IP with your actual EC2 public IP address

On EC2 (Cloud Server)

Terminal 1 - Start Zenoh Router:

zenohd --listen tcp/0.0.0.0:7447

Terminal 2 - Start Dora Coordinator:

cd ~/dora-distributed
dora coordinator

Terminal 3 - Start Cloud Daemon:

cd ~/dora-distributed
export ZENOH_CONFIG=zenoh_config.json5
dora daemon --machine-id cloud

You should see:

INFO dora_daemon::coordinator: Connected to dora-coordinator at 127.0.0.1:53290

On Your PC/Mac (Edge Device)

Terminal 1 - Start Edge Daemon:

cd /path/to/dora-examples/examples/python-distributed-zenoh
export ZENOH_CONFIG=zenoh_config.json5
dora daemon --coordinator-addr YOUR_EC2_IP --machine-id edge1

You should see:

INFO dora_daemon::coordinator: Connected to dora-coordinator at YOUR_EC2_IP:53290

Start the Dataflow (From EC2)

Terminal 4 on EC2:

cd ~/dora-distributed

# Verify both daemons are connected
dora list

# Build the dataflow
dora build dataflow_distributed.yml

# Start the dataflow (controls both machines)
dora start dataflow_distributed.yml

Expected Output

On EC2 (Terminal 4) - Cloud Aggregator:

dataflow start triggered: 019a5637-3863-7f52-aac5-df670c2a7132
attaching to dataflow (use `--detach` to run in background)

cloud-aggregator on daemon `cloud`: INFO   daemon    node is ready
cloud-aggregator on daemon `cloud`: stdout    === Cloud Aggregator Node ===
cloud-aggregator on daemon `cloud`: stdout    Receiving data via Dora's distributed Zenoh routing
cloud-aggregator on daemon `cloud`: stdout    
cloud-aggregator on daemon `cloud`: stdout    Cloud aggregator started! Press Ctrl+C to stop
cloud-aggregator on daemon `cloud`: stdout    
cloud-aggregator on daemon `cloud`: stdout    >> [00001] Received from sensor_temp: temperature=23.45celsius
cloud-aggregator on daemon `cloud`: stdout    >> [00002] Received from sensor_temp: temperature=24.12celsius
cloud-aggregator on daemon `cloud`: stdout    >> [00003] Received from sensor_temp: temperature=22.89celsius
cloud-aggregator on daemon `cloud`: stdout    
cloud-aggregator on daemon `cloud`: stdout    ======================================================================
cloud-aggregator on daemon `cloud`: stdout    Cloud Aggregator Summary - Total Messages: 10
cloud-aggregator on daemon `cloud`: stdout    Active Sensors: 1
cloud-aggregator on daemon `cloud`: stdout    ======================================================================
cloud-aggregator on daemon `cloud`: stdout    🟢 ACTIVE | sensor_temp     | temperature  |   24.12 celsius  | Count:    10
cloud-aggregator on daemon `cloud`: stdout    ======================================================================

On Your PC/Mac (Terminal 1) - Edge Daemon:

INFO dora_daemon::coordinator: Connected to dora-coordinator at YOUR_EC2_IP:53290
WARN run_inner: dora_daemon: Daemon took 158ms for handling event

The sensor node runs silently in the background, sending temperature data every second through the Zenoh router to EC2.

Architecture Flow

Your PC/Mac (edge1)          EC2 Cloud (YOUR_EC2_IP)
┌─────────────────┐           ┌─────────────────────────┐
│ sensor-temp     │           │ zenohd :7447 (router)   │
│ (publishes)     │──Zenoh──▶│                         │
│                 │           │ dora coordinator :53290 │
│ dora daemon     │◀─Control─│                         │
│ (edge1)         │           │ dora daemon (cloud)     │
└─────────────────┘           │                         │
                              │ cloud-aggregator        │
                              │ (receives & aggregates) │
                              └─────────────────────────┘

Troubleshooting Cross-Network Setup

"Connection refused" when starting daemon:

Verify EC2 public IP is correct in --coordinator-addr
Check EC2 security group allows TCP port 53290 (coordinator)
Ensure coordinator is running on EC2

"Operation not supported (os error 45)" on Mac:

Update dataflow_distributed.yml with correct Mac path in working_dir
Mac uses /Users/..., not /home/...

"Address already in use" for port 7447:

Update zenoh_config.json5 with mode: "client" and listen: { endpoints: [] }
Only the zenohd router should listen on 7447

Zenoh connection issues:

Verify EC2 security group allows TCP port 7447 (Zenoh router) and TCP port 53290 (coordinator)
Test Zenoh connectivity: z_pub --connect tcp/<EC2_IP>:7447 --key test
On EC2, verify services are listening: netstat -tuln | grep -E '7447|53290'

Expected Output

Sensor Node (edge1):

=== Sensor Node: sensor_temp ===
Type: temperature
Sensor node sensor_temp started!
Data will be automatically routed via Dora's Zenoh integration
Press Ctrl+C to stop

[0001] Sent: temperature=23.45celsius
[0002] Sent: temperature=24.12celsius
[0003] Sent: temperature=22.89celsius
...

Cloud Aggregator:

=== Cloud Aggregator Node ===
Receiving data via Dora's distributed Zenoh routing

Cloud aggregator started! Press Ctrl+C to stop

>> [00001] Received from sensor_temp: temperature=23.45celsius
>> [00002] Received from sensor_humidity: humidity=65.32percent
>> [00003] Received from sensor_temp: temperature=24.12celsius
>> [00004] Received from sensor_pressure: pressure=1013.45hPa

======================================================================
Cloud Aggregator Summary - Total Messages: 15
Active Sensors: 3
======================================================================
🟢 ACTIVE | sensor_temp     | temperature  |   24.12 celsius  | Count:     5
🟢 ACTIVE | sensor_humidity | humidity     |   65.32 percent  | Count:     4
🟢 ACTIVE | sensor_pressure | pressure     | 1013.45 hPa      | Count:     3
======================================================================

How Dora's Distribution Works

Coordinator: Central manager that tracks all daemons and dataflows
Daemons: One per machine, identified by --machine-id
Machine Assignment: Nodes use _unstable_deploy.machine in YAML
Automatic Routing:
- Same machine: Direct communication
- Same network: Zenoh multicast auto-discovery
- Different networks: Zenoh routers (configured via ZENOH_CONFIG)

Deployment Checklist

Copy Python scripts to each target machine
Install dora-rs on all machines: pip install dora-rs
Start coordinator on accessible server
Start daemon on each machine with unique --machine-id
Machine IDs in YAML must match daemon --machine-id
For cross-network: Set up Zenoh routers and ZENOH_CONFIG

Troubleshooting

"No such file or directory" error:

Ensure working_dir is set in _unstable_deploy section
Verify Python scripts exist at the working directory path
For distributed setups, copy scripts to the same path on all machines

Nodes not connecting:

Check coordinator IP is accessible from all machines
Verify machine IDs in YAML match daemon --machine-id
Check firewall allows Zenoh ports (default: UDP 7447)

No data received:

Verify nodes are running: dora list
Check logs: dora logs sensor-temp, dora logs cloud-aggregator
Ensure all Python scripts are present on their target machines

Cross-network issues:

Verify Zenoh router (zenohd) is running
Check ZENOH_CONFIG points to correct router
Test Zenoh connectivity: z_ping

Zenoh warnings (tcp/[::]:5456 already in use):

Safe to ignore when running multiple daemons locally
Daemons will automatically retry with alternative configuration

Zenoh Topic Structure

Dora uses this Zenoh topic naming:

dora/{network_id}/{dataflow_id}/output/{node_id}/{output_id}

Example:

dora/default/019a41c7-0718-75ca-82fe-66ad9b289b5b/output/sensor-temp/data

You can monitor these topics directly with Zenoh tools:

# Subscribe to all dora topics
z_sub 'dora/**'

Testing Network Connectivity

Before running dora-zenoh, verify Zenoh connectivity works:

Terminal A (on edge or EC2):

z_sub --connect tcp/YOUR_EC2_IP:7447 --key demo/test

Terminal B (on another machine):

echo "🔥 connected!" | z_pub --connect tcp/YOUR_EC2_IP:7447 --key demo/test

Expected output on Terminal A:

Received (key='demo/test'): 🔥 connected!

If this works, your network is configured correctly. If dora still has issues, check the dora-specific troubleshooting above.

Python Distributed Dora Sensor Network

Key Features

Architecture

Dataflow Configuration

Installation

On All Machines

AWS EC2 Security Group Configuration (For Cross-Network)

Running on ONE Machine (Simulation)

Manual Steps

Running on real distribution

Same Network (WiFi/LAN)

Different Networks (Cross-Network) - AWS EC2 Example

Prerequisites

Setup Zenoh Config

On EC2 (Cloud Server)

On Your PC/Mac (Edge Device)

Start the Dataflow (From EC2)

Expected Output

Architecture Flow

Troubleshooting Cross-Network Setup

Expected Output

How Dora's Distribution Works

Deployment Checklist

Troubleshooting

Zenoh Topic Structure

Testing Network Connectivity

References