popov/neuron-tai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

relay over ws supossedly working

Browse Source
This commit is contained in:
Dobromir Popov
2026-06-30 21:06:39 +02:00
parent d0307fcc84
commit 61074a8fe8
6 changed files with 429 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

105
QUICKSTART.md
View File

@@ -135,6 +135,13 @@ Via public hostname with relay (works from behind NAT, WSL2, 5G — no `--advert
.\.venv\Scripts\meshnet-node.exe start --tracker https://ai.neuron.d-popov.com --model Qwen/Qwen2.5-0.5B-Instruct
```
WSL (same relay path — no `--advertise-host`):
```bash
.venv/bin/meshnet-node start --tracker https://ai.neuron.d-popov.com --model Qwen/Qwen2.5-0.5B-Instruct
.venv/bin/meshnet-node start --tracker http://192.168.0.179:8081 --model Qwen/Qwen2.5-0.5B-Instruct
```
`--host 0.0.0.0` binds the node to all Windows interfaces. `--advertise-host`
is what the tracker gives to other nodes for direct connections; omit it when using
the relay path since all traffic flows through the relay tunnel instead.
@@ -176,26 +183,114 @@ Client → HTTPS → ai.neuron.d-popov.com (nginx)
└─ /rpc/* → meshnet-relay :8765 (caller opens WS per hop)
```
### Start the relay and tracker (server side)
### Nginx Proxy Manager (Docker)
Use **one** proxy host for the domain. Do not create a second host for the same
domain to reach another port — path routing is done with **Custom locations** on
that same host.
**1. Details tab** (default `/` → tracker)
| Field | Value |
|-------|--------|
| Domain Names | `ai.neuron.d-popov.com` |
| Scheme | `http` |
| Forward Hostname / IP | LAN IP of the tracker machine (e.g. `192.168.0.179`) |
| Forward Port | `8081` |
| Websockets Support | ON |
This serves `/v1/network/map`, `/v1/chat/completions`, and the rest of the tracker API.
**2. Custom locations tab** (sub-paths → relay)
The Custom locations form has **no separate Websockets toggle** — only location,
scheme, forward host, optional sub-folder path, and port. Add **two** locations
(both pointing at the relay process on port `8765`). Leave **“Add a path for
sub-folder forwarding”** empty so the full URI reaches the relay
(`/ws`, `/rpc/<peer_id>`).
Location A — persistent node connections:
| Field | Value |
|-------|--------|
| Define location | `/ws` |
| Scheme | `http` |
| Forward Hostname / IP | `192.168.0.179` |
| Forward Port | `8765` |
| Sub-folder path | *(leave empty)* |
Location B — per-hop RPC:
| Field | Value |
|-------|--------|
| Define location | `/rpc` |
| Scheme | `http` |
| Forward Hostname / IP | `192.168.0.179` |
| Forward Port | `8765` |
| Sub-folder path | *(leave empty)* |
Nginx matches the longer prefixes first: `/ws` and `/rpc/…` go to relay; everything
else stays on `8081`.
**3. SSL tab**
Use your existing Let’s Encrypt certificate (unchanged).
**4. Advanced tab** (only if WebSocket upgrade fails on `/ws` or `/rpc`)
Custom locations do not expose a Websockets checkbox. If nodes show
`Relay configured but not connected yet` while `/v1/network/map` works, add this
snippet on the **proxy host** Advanced tab:
```nginx
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection $http_connection;
proxy_read_timeout 3600s;
proxy_send_timeout 3600s;
```
**5. Verify routing**
```bash
# Terminal 1 — relay (WebSocket hub)
# Tracker (8081 via default location)
curl -s https://ai.neuron.d-popov.com/v1/network/map | python3 -m json.tool
# Relay paths should not 502/404 from the tracker — check response headers/status
curl -sI https://ai.neuron.d-popov.com/ws
curl -sI https://ai.neuron.d-popov.com/rpc/test-peer
```
After NPM is correct, start relay and tracker on the LAN machine:
```bash
# Terminal 1 — relay
.venv/bin/meshnet-relay --host 0.0.0.0 --port 8765
# Terminal 2 — tracker (advertises relay URL to nodes)
# Terminal 2 — tracker (advertises relay to nodes)
.venv/bin/meshnet-tracker start \
--host 0.0.0.0 \
--port 8081 \
--relay-url wss://ai.neuron.d-popov.com/ws
```
Nodes using `https://ai.neuron.d-popov.com` should then log:
```text
Relay advertised by tracker — using outbound tunnel wss://ai.neuron.d-popov.com/ws
Relay connected — wss://ai.neuron.d-popov.com/rpc/<peer_id>
```
The `--relay-url` flag embeds the relay address in `/v1/network/map`. Every node
queries that endpoint on startup and auto-connects if a relay URL is present.
### Start a node (any machine, any network)
No `--advertise-host` needed. The node discovers the relay URL from the tracker
and opens a persistent outbound WebSocket:
No `--advertise-host`, firewall rule, port forwarding, relay URL, or peer URL is
needed on the node. The public tracker is the only bootstrap URL the user types.
The node queries the tracker for `/v1/network/map`, discovers the relay URL, and
opens a persistent outbound WebSocket. If the relay connection drops, the node
keeps retrying it.
```bash
.venv/bin/meshnet-node start \

40
packages/node/meshnet_node/startup.py
View File

@@ -35,13 +35,32 @@ def _get_json(url: str, timeout: float = 10.0) -> dict:
return json.loads(r.read())
def _infer_relay_url_from_tracker(tracker_url: str) -> str | None:
"""Infer relay WebSocket URL from a public HTTPS tracker origin.
Public deployments colocate relay at /ws on the same host as the tracker API
(see QUICKSTART nginx layout). Local LAN trackers use a separate relay port
and must advertise relay_url explicitly via /v1/network/map.
"""
parsed = urllib.parse.urlparse(tracker_url)
if parsed.scheme != "https":
return None
host = parsed.hostname
if not host or host in ("127.0.0.1", "localhost"):
return None
return f"wss://{parsed.netloc}/ws"
def _discover_relay_url(tracker_url: str) -> str | None:
relay_url: str | None = None
try:
network_map = _get_json(f"{tracker_url}/v1/network/map", timeout=5.0)
raw = network_map.get("relay_url")
if isinstance(raw, str) and raw:
relay_url = raw
except Exception:
return None
relay_url = network_map.get("relay_url")
return relay_url if isinstance(relay_url, str) and relay_url else None
pass
return relay_url or _infer_relay_url_from_tracker(tracker_url)
def _start_relay_bridge_if_available(
@@ -49,8 +68,9 @@ def _start_relay_bridge_if_available(
wallet_address: str,
local_base_url: str,
advertised_endpoint: str,
relay_url: str | None = None,
) -> tuple[RelayHttpBridge | None, dict]:
relay_url = _discover_relay_url(tracker_url)
relay_url = relay_url or _discover_relay_url(tracker_url)
if not relay_url:
return None, {}
peer_id = peer_id_from_wallet(wallet_address)
@@ -222,7 +242,8 @@ def _warn_virtual_network_ip(ip: str | None) -> None:
f" This range is used by Docker, WSL2, and virtual machines and is\n"
f" NOT reachable from other physical machines on your LAN.\n"
f" Cross-host pipeline hops WILL time out.\n"
f" Fix: pass --advertise-host <your-LAN-ip> (e.g. 192.168.x.x).\n",
f" Fix: use a public tracker with relay (wss://…/ws), or pass\n"
f" --advertise-host <your-LAN-ip> (e.g. 192.168.x.x).\n",
flush=True,
)
except Exception:
@@ -260,6 +281,7 @@ def run_startup(
"""
tracker_url = tracker_url.rstrip("/")
relay_url = _discover_relay_url(tracker_url)
# 1. Hardware detection
if advertise_host is None and host == "0.0.0.0":
@@ -275,7 +297,10 @@ def run_startup(
except Exception:
advertise_host = socket.getfqdn()
_warn_virtual_network_ip(advertise_host)
if relay_url:
print(f"Relay advertised by tracker — using outbound tunnel {relay_url}", flush=True)
else:
_warn_virtual_network_ip(advertise_host)
print("Detecting hardware...", flush=True)
hw = detect_hardware()
@@ -361,6 +386,7 @@ def run_startup(
address,
local_base_url,
endpoint,
relay_url=relay_url,
)
_attach_relay_bridge(node, relay_bridge)
# Register with tracker so other nodes can auto-join this model.
@@ -451,6 +477,7 @@ def run_startup(
address,
local_base_url,
endpoint,
relay_url=relay_url,
)
_attach_relay_bridge(node, relay_bridge)
auto_reg_payload = {
@@ -548,6 +575,7 @@ def run_startup(
address,
local_base_url,
endpoint,
relay_url=relay_url,
)
_attach_relay_bridge(node, relay_bridge)

5
packages/tracker/meshnet_tracker/cli.py
View File

@@ -4,7 +4,7 @@ import argparse
import sys
import time
from .server import TrackerServer
from .server import TrackerServer, derive_relay_url_from_public_tracker_url
def main() -> None:
@@ -52,6 +52,7 @@ def main() -> None:
if args.command in {None, "start"}:
cluster_peers = [u.strip() for u in args.cluster_peers.split(",") if u.strip()]
relay_url = args.relay_url or derive_relay_url_from_public_tracker_url(args.self_url)
server = TrackerServer(
host=args.host,
port=args.port,
@@ -59,7 +60,7 @@ def main() -> None:
cluster_peers=cluster_peers or None,
cluster_self_url=args.self_url,
stats_db=getattr(args, "stats_db", None),
relay_url=args.relay_url,
relay_url=relay_url,
)
port = server.start()
print(f"meshnet-tracker listening on http://{args.host}:{port}", flush=True)

65
packages/tracker/meshnet_tracker/server.py
View File

@@ -37,6 +37,19 @@ from .gossip import NodeGossip
from .raft import RaftNode
def derive_relay_url_from_public_tracker_url(url: str | None) -> str | None:
"""Return wss://host/ws when url is a public HTTPS tracker origin."""
if not url:
return None
parsed = urllib.parse.urlparse(url)
if parsed.scheme != "https":
return None
host = parsed.hostname
if not host or host in ("127.0.0.1", "localhost"):
return None
return f"wss://{parsed.netloc}/ws"
DEFAULT_MODEL_PRESETS: dict[str, dict] = {
"stub-model": {
"layers_start": 0,
@@ -1228,6 +1241,33 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
if auth:
req.add_header("Authorization", auth)
relay_headers = {
"Content-Type": "application/json",
"X-Meshnet-Route": downstream_urls,
**({"Authorization": auth} if auth else {}),
}
if node.relay_addr:
print(
f"[tracker] proxy via relay {request_id}: {node.relay_addr} "
f"(direct endpoint {target_url})",
flush=True,
)
relayed = _relay_http_request(
node.relay_addr,
path="/v1/chat/completions",
body=raw_body,
headers=relay_headers,
)
if relayed is not None:
self._send_relayed_response(relayed)
return
print(
f"[tracker] relay proxy failed {request_id}: {node.relay_addr}; "
f"trying direct {target_url}",
flush=True,
)
try:
upstream = urllib.request.urlopen(req, timeout=300.0)
print(f"[tracker] proxy connected {request_id}: {target_url}", flush=True)
@@ -1247,23 +1287,11 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
if node.relay_addr:
print(
f"[tracker] direct proxy failed {request_id}: {target_url}: {exc}; "
f"trying relay {node.relay_addr}",
f"relay already attempted for {node.relay_addr}",
flush=True,
)
relayed = _relay_http_request(
node.relay_addr,
path="/v1/chat/completions",
body=raw_body,
headers={
"Content-Type": "application/json",
"X-Meshnet-Route": downstream_urls,
**({"Authorization": auth} if auth else {}),
},
)
if relayed is not None:
self._send_relayed_response(relayed)
return
print(f"[tracker] proxy failed {request_id}: {target_url}: {exc}", flush=True)
else:
print(f"[tracker] proxy failed {request_id}: {target_url}: {exc}", flush=True)
self._send_json(503, {"error": {
"message": f"upstream node unreachable: {exc}",
"type": "service_unavailable",
@@ -2040,6 +2068,11 @@ class TrackerServer:
if self._server is not None:
raise RuntimeError("TrackerServer is already running")
effective_relay_url = (
self._relay_url
or derive_relay_url_from_public_tracker_url(self._cluster_self_url)
)
# Start HTTP server first so we know our port
self._server = _TrackerHTTPServer(
(self._host, self._requested_port),
@@ -2050,7 +2083,7 @@ class TrackerServer:
self._model_presets,
self._contracts,
self._minimum_stake,
relay_url=self._relay_url,
relay_url=effective_relay_url,
stats=self._stats,
)
self.port = self._server.server_address[1]

66
tests/test_gossip_and_relay.py
View File

@@ -350,6 +350,51 @@ def test_relay_rpc_round_trips_http_request_to_peer():
assert json.loads(response["body"]) == {"ok": True, "path": "/v1/chat/completions"}
def test_node_relay_bridge_reconnects_after_failed_connection(monkeypatch):
"""Node-side relay bridge keeps retrying its outbound WebSocket connection."""
import websockets.sync.client as wsc # type: ignore[import]
from meshnet_node.relay_bridge import RelayHttpBridge
attempts = []
class FakeWebSocket:
def __enter__(self):
return self
def __exit__(self, exc_type, exc, tb):
return False
def send(self, _message):
pass
def recv(self, timeout=1):
raise TimeoutError
def fake_connect(url, open_timeout=5):
attempts.append((url, open_timeout))
if len(attempts) == 1:
raise OSError("temporary relay outage")
return FakeWebSocket()
monkeypatch.setattr(wsc, "connect", fake_connect)
bridge = RelayHttpBridge(
relay_url="ws://relay.example/ws",
peer_id="peer-reconnect",
local_base_url="http://127.0.0.1:8001",
advertised_addr="http://127.0.0.1:8001",
reconnect_interval=0.01,
)
try:
info = bridge.start()
assert info.relay_addr == "ws://relay.example/rpc/peer-reconnect"
assert bridge.wait_connected(timeout=1.0)
finally:
bridge.stop()
assert len(attempts) >= 2
# ---------------------------------------------------------------------------
# Tracker gossip fields tests
# ---------------------------------------------------------------------------
@@ -389,6 +434,27 @@ def _start_tracker_and_register(extra_fields: dict) -> dict:
return resp
def test_tracker_derives_relay_url_from_public_self_url():
from meshnet_tracker.server import TrackerServer, derive_relay_url_from_public_tracker_url
assert derive_relay_url_from_public_tracker_url("https://ai.neuron.d-popov.com") == (
"wss://ai.neuron.d-popov.com/ws"
)
assert derive_relay_url_from_public_tracker_url("http://127.0.0.1:8081") is None
tracker = TrackerServer(cluster_self_url="https://ai.neuron.d-popov.com")
port = tracker.start()
try:
import json as _json
import urllib.request
with urllib.request.urlopen(f"http://127.0.0.1:{port}/v1/network/map", timeout=5) as resp:
body = _json.loads(resp.read())
assert body["relay_url"] == "wss://ai.neuron.d-popov.com/ws"
finally:
tracker.stop()
def test_tracker_accepts_relay_addr_in_registration():
resp = _start_tracker_and_register({
"relay_addr": "ws://relay.meshnet.ai:8765/relay/abc123",

180
tests/test_node_startup.py
View File

@@ -11,7 +11,11 @@ import pytest
from meshnet_node.downloader import download_shard, write_shard_archive
from meshnet_node.hardware import detect_hardware
from meshnet_node.startup import _probationary_status_line, run_startup
from meshnet_node.startup import (
_infer_relay_url_from_tracker,
_probationary_status_line,
run_startup,
)
from meshnet_node.wallet import _b58encode, load_or_create_wallet
from meshnet_contracts import LocalSolanaContracts
from meshnet_tracker.server import TrackerServer
@@ -347,6 +351,94 @@ def test_tracker_assign_lists_peers_for_same_model_shard():
tracker.stop()
def test_infer_relay_url_from_public_https_tracker():
assert _infer_relay_url_from_tracker("https://ai.neuron.d-popov.com") == (
"wss://ai.neuron.d-popov.com/ws"
)
assert _infer_relay_url_from_tracker("https://ai.neuron.d-popov.com/v1/network/map") == (
"wss://ai.neuron.d-popov.com/ws"
)
assert _infer_relay_url_from_tracker("http://192.168.0.179:8081") is None
assert _infer_relay_url_from_tracker("http://127.0.0.1:8081") is None
def test_public_https_tracker_infers_relay_when_network_map_omits_relay_url(
tmp_path,
monkeypatch,
capsys,
):
"""Nodes bootstrap relay from the tracker origin when map relay_url is null."""
import meshnet_node.startup as startup_mod
class FakeBackend:
total_layers = 24
class FakeTorchNodeServer:
def __init__(self, **kwargs):
self.backend = FakeBackend()
self.port = None
self.chat_completion_count = 0
self.total_requests = 0
self.failed_requests = 0
self.queue_depth = 0
def start(self):
self.port = 8001
return self.port
def stop(self):
pass
class FakeRelayHttpBridge:
def __init__(self, relay_url, peer_id, local_base_url, advertised_addr):
self.relay_url = relay_url
self.peer_id = peer_id
@property
def relay_addr(self):
return f"{self.relay_url.replace('/ws', '')}/rpc/{self.peer_id}"
def start(self):
return types.SimpleNamespace(peer_id=self.peer_id, relay_addr=self.relay_addr)
def wait_connected(self, timeout=5.0):
return True
def stop(self):
pass
monkeypatch.setattr(
startup_mod,
"detect_hardware",
lambda: {"device": "cpu", "gpu_name": None, "vram_mb": 0},
)
monkeypatch.setattr(startup_mod, "_detect_num_layers", lambda _model_id: 24)
monkeypatch.setattr(startup_mod, "TorchNodeServer", FakeTorchNodeServer)
monkeypatch.setattr(startup_mod, "RelayHttpBridge", FakeRelayHttpBridge)
monkeypatch.setattr(
startup_mod,
"_get_json",
lambda url, timeout=10.0: {"relay_url": None, "nodes": []},
)
tracker_url = "https://ai.neuron.d-popov.com"
node = run_startup(
tracker_url=tracker_url,
model_id="Qwen/Qwen2.5-0.5B-Instruct",
advertise_host="172.29.104.23",
wallet_path=tmp_path / "wallet.json",
)
try:
pass
finally:
node.stop()
output = capsys.readouterr().out
assert "Relay advertised by tracker" in output
assert "wss://ai.neuron.d-popov.com/ws" in output
assert "Cross-host pipeline hops WILL time out" not in output
def test_real_model_startup_summary_shows_total_layers(tmp_path, monkeypatch, capsys):
"""Real-model startup summary prints the shard range plus total model layers."""
import meshnet_node.startup as startup_mod
@@ -394,6 +486,7 @@ def test_real_model_startup_summary_shows_total_layers(tmp_path, monkeypatch, ca
def test_public_tracker_model_node_registers_relay_metadata_from_tracker_url_only(
tmp_path,
monkeypatch,
capsys,
):
"""A node only needs the public tracker URL to discover relay metadata and register."""
import meshnet_node.startup as startup_mod
@@ -457,7 +550,6 @@ def test_public_tracker_model_node_registers_relay_metadata_from_tracker_url_onl
node = run_startup(
tracker_url=tracker_url,
model_id="Qwen/Qwen2.5-0.5B-Instruct",
advertise_host="203.0.113.10",
wallet_path=tmp_path / "wallet.json",
)
try:
@@ -471,9 +563,91 @@ def test_public_tracker_model_node_registers_relay_metadata_from_tracker_url_onl
assert len(network_map["nodes"]) == 1
registered = network_map["nodes"][0]
assert registered["hf_repo"] == "Qwen/Qwen2.5-0.5B-Instruct"
assert registered["endpoint"] == "http://203.0.113.10:8001"
assert registered["endpoint"].startswith("http://")
assert registered["endpoint"].endswith(":8001")
assert registered["relay_addr"].startswith("ws://public-relay.example/rpc/")
assert registered["peer_id"]
output = capsys.readouterr().out
assert "Relay advertised by tracker" in output
assert "Cross-host pipeline hops WILL time out" not in output
def test_public_tracker_relay_suppresses_virtual_ip_warning(
tmp_path,
monkeypatch,
capsys,
):
"""A WSL/Docker endpoint is acceptable when the tracker advertises relay RPC."""
import meshnet_node.startup as startup_mod
class FakeBackend:
total_layers = 24
class FakeTorchNodeServer:
def __init__(self, **kwargs):
self.backend = FakeBackend()
self.port = None
self.chat_completion_count = 0
self.total_requests = 0
self.failed_requests = 0
self.queue_depth = 0
def start(self):
self.port = 8001
return self.port
def stop(self):
pass
class FakeRelayHttpBridge:
def __init__(self, relay_url, peer_id, local_base_url, advertised_addr):
self.relay_url = relay_url
self.peer_id = peer_id
@property
def relay_addr(self):
return f"ws://public-relay.example/rpc/{self.peer_id}"
def start(self):
return types.SimpleNamespace(peer_id=self.peer_id, relay_addr=self.relay_addr)
def wait_connected(self, timeout=5.0):
return True
def stop(self):
pass
monkeypatch.setattr(
startup_mod,
"detect_hardware",
lambda: {"device": "cpu", "gpu_name": None, "vram_mb": 0},
)
monkeypatch.setattr(startup_mod, "_detect_num_layers", lambda _model_id: 24)
monkeypatch.setattr(startup_mod, "TorchNodeServer", FakeTorchNodeServer)
monkeypatch.setattr(startup_mod, "RelayHttpBridge", FakeRelayHttpBridge)
tracker = TrackerServer(relay_url="ws://public-relay.example/ws")
tracker_port = tracker.start()
tracker_url = f"http://127.0.0.1:{tracker_port}"
try:
node = run_startup(
tracker_url=tracker_url,
model_id="Qwen/Qwen2.5-0.5B-Instruct",
advertise_host="172.29.104.23",
wallet_path=tmp_path / "wallet.json",
)
try:
network_map = _get_json(f"{tracker_url}/v1/network/map")
finally:
node.stop()
finally:
tracker.stop()
assert network_map["nodes"][0]["endpoint"] == "http://172.29.104.23:8001"
assert network_map["nodes"][0]["relay_addr"].startswith("ws://public-relay.example/rpc/")
output = capsys.readouterr().out
assert "Relay advertised by tracker" in output
assert "Cross-host pipeline hops WILL time out" not in output
def test_later_node_auto_joins_existing_public_hf_model_with_only_tracker_url(