Assistance Required: Handling "Unknown Server" Responses in aiocoap-Based LwM2M Server Implementation

[Xoein]

Hello,

I hope that I'm addressing my request in the good place

I am currently developing a Lightweight M2M (LwM2M) server using the aiocoap library and have encountered an issue where client devices respond with "Unknown Server" to certain requests. I am seeking guidance on how to configure aiocoap to ensure consistent communication between the server and clients.

Context:

• Server Setup: The server is implemented using aiocoap's Context.create_server_context, bound to a specific IP address and port (e.g., 127.0.0.3:5683). It handles client registrations and is intended to send requests back to clients using the same address and port.
• Issue Encountered: While the server successfully receives registration requests from clients, attempts to send CoAP GET requests to clients result in "Unknown Server" responses. It appears that the outgoing requests may not be originating from the same address and port to which the clients have registered, leading to this issue.

Code Snippet Illustrating the Issue:

import asyncio
from aiocoap import Context, Message, GET
from aiocoap.resource import Site
from coap_resources import RegisterResource
import config

class CoAPServer:
    def __init__(self):
        self.coap_server_context = None
        self.root = None
        self.server = '127.0.0.3'
        self.server_port = 5683

    async def start_coap_server(self):
        self.root = Site()
        self.root.add_resource(('rd',), RegisterResource(config.socketio, self.root, self))
        bind_address = (self.server, self.server_port)

        try:
            self.coap_server_context = await Context.create_server_context(self.root, bind=bind_address)
            print(f"CoAP server is running on coap://{bind_address[0]}:{bind_address[1]}")
        except OSError as e:
            print(f"Failed to bind server to {bind_address}: {e}")
            return

        await asyncio.get_running_loop().create_future()

    async def send_coap_request(self, uri):
        try:
            request = Message(code=GET, uri=uri)
            response = await self.coap_server_context.request(request).response
            print(f"Received response: {response.code} - {response.payload.decode('utf-8')}")
            return response.payload.decode("utf-8")
        except Exception as e:
            print(f"Error during GET request to {uri}: {e}")
            raise

# Initialization and server start logic...

Steps Taken:

1. Binding Server Context: The server context is created with a specific bind address to ensure it listens on the desired IP and port. 2. Sending Requests: GET requests to clients are sent using the same server context to maintain consistent source address and port.

Request for Assistance:

I would appreciate guidance on the following:

1. Ensuring Consistent Source Address and Port: How can I configure aiocoap to guarantee that outgoing requests to clients originate from the same address and port to which clients have registered, thereby preventing "Unknown Server" responses?

2. Best Practices for LwM2M Implementations: Are there recommended configurations or patterns within aiocoap for implementing LwM2M servers that require consistent communication channels with clients?

Your expertise and any insights you can provide on resolving this issue would be greatly appreciated.

Thank you for your assistance.

Best regards, Sameh

Address/port used in registration:

Address/port of get request:

[chrysn]

Was this reported somewhere else too? I clearly remember having discussed the 127.0.0.3 issue before, but don't find where. Anyhow:

The surprising part here is not so much the port but the IP address that does not match. Can you reproduce this behavior on IPv6? (There are no multiple loopback addresses on IPv6, but you can for example use a local ULA, or just bind to a different port on ::1).