fix(us-020): silence BrokenPipeError in tracker _send_json; deterministic node IDs; HF model coverage

- Wrap wfile.write in _TrackerHandler._send_json with except BrokenPipeError
- Replace uuid4 node IDs with deterministic SHA-256 hash of endpoint+model+shards
  so nodes keep the same ID on re-registration after tracker restart
- /v1/models now lists HF-repo models (not just preset models)
- /v1/coverage/{model} now resolves HF repos, not just preset names
- /v1/route response includes node_id alongside endpoint
- startup.py exposes tracker_node_id on node object and prints it in dashboard

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Dobromir Popov
2026-06-30 12:26:50 +03:00
parent 4a803377dc
commit 2e1e0ae172
4 changed files with 256 additions and 16 deletions

View File

@@ -227,12 +227,15 @@ def run_startup(
"score": 1.0,
"tracker_mode": (shard_start == 0),
}
tracker_node_id: str | None = None
try:
reg_resp = _post_json(f"{tracker_url}/v1/nodes/register", reg_payload)
node_id = reg_resp.get("node_id", "?")
print(f" Registered with tracker — node ID: {node_id}", flush=True)
_start_heartbeat(tracker_url, node_id, reg_payload)
tracker_node_id = str(reg_resp.get("node_id") or "?")
setattr(node, "tracker_node_id", tracker_node_id)
print(f" Registered with tracker — node ID: {tracker_node_id}", flush=True)
_start_heartbeat(tracker_url, tracker_node_id, reg_payload)
except Exception as exc:
setattr(node, "tracker_node_id", None)
print(f" Warning: tracker registration failed: {exc}", flush=True)
print(
@@ -243,6 +246,7 @@ def run_startup(
f" Shard: {shard_label}\n"
f" Quantization: {quantization}\n"
f" Endpoint: {endpoint}\n"
f" Node ID: {tracker_node_id or 'unregistered'}\n"
f" Hardware: {device.upper()}\n"
f"{'=' * 32}",
flush=True,
@@ -298,12 +302,15 @@ def run_startup(
"score": 1.0,
"tracker_mode": (assigned_shard_start == 0),
}
tracker_node_id = None
try:
reg_resp = _post_json(f"{tracker_url}/v1/nodes/register", auto_reg_payload)
node_id = reg_resp.get("node_id", "?")
print(f" Registered with tracker — node ID: {node_id}", flush=True)
_start_heartbeat(tracker_url, node_id, auto_reg_payload)
tracker_node_id = str(reg_resp.get("node_id") or "?")
setattr(node, "tracker_node_id", tracker_node_id)
print(f" Registered with tracker — node ID: {tracker_node_id}", flush=True)
_start_heartbeat(tracker_url, tracker_node_id, auto_reg_payload)
except Exception as exc:
setattr(node, "tracker_node_id", None)
print(f" Warning: tracker registration failed: {exc}", flush=True)
shard_count = assigned_shard_end - assigned_shard_start + 1
print(
@@ -315,6 +322,7 @@ def run_startup(
f"({shard_count} of {assigned_num_layers})\n"
f" Quantization: {quantization}\n"
f" Endpoint: {endpoint}\n"
f" Node ID: {tracker_node_id or 'unregistered'}\n"
f" Hardware: {device.upper()}\n"
f"{'=' * 32}",
flush=True,
@@ -385,7 +393,8 @@ def run_startup(
"score": 1.0,
},
)
node_id: str = reg_resp["node_id"]
node_id = str(reg_resp["node_id"])
setattr(node, "tracker_node_id", node_id)
except Exception:
node.stop()
raise

View File

@@ -22,11 +22,11 @@ HTTP API contract:
"""
import http.server
import hashlib
import json
import threading
import time
import urllib.parse
import uuid
from typing import Any
from .gossip import NodeGossip
@@ -230,6 +230,27 @@ def _coverage_gaps(coverage: list[dict]) -> list[tuple[int, int]]:
]
def _nodes_and_bounds_for_model(
server: "_TrackerHTTPServer",
model: str,
) -> tuple[list[_NodeEntry], int, int] | None:
preset = server.model_presets.get(model)
if preset is not None:
required_start, required_end = _preset_layer_bounds(preset)
return [node for node in server.registry.values() if node.model == model], required_start, required_end
nodes = [
node for node in server.registry.values()
if (node.hf_repo == model or node.model == model)
and node.shard_start is not None
and node.shard_end is not None
and node.num_layers is not None
]
if not nodes:
return None
return nodes, 0, max(node.num_layers for node in nodes) - 1
def _load_directive(node: _NodeEntry, model: str, start: int, end: int, quantization: str) -> dict:
return {
"action": "LOAD_SHARD",
@@ -369,6 +390,27 @@ def _registration_ban_error(contracts: Any | None, wallet_address: str | None) -
return None
def _node_id_for_registration(
endpoint: str,
model: str,
wallet_address: str | None,
shard_start: int | None,
shard_end: int | None,
hf_repo: str | None,
) -> str:
wallet_prefix = wallet_address[:8] if wallet_address else "anon"
stable_key = "|".join([
wallet_address or "",
endpoint.rstrip("/"),
model,
hf_repo or "",
"" if shard_start is None else str(shard_start),
"" if shard_end is None else str(shard_end),
])
digest = hashlib.sha256(stable_key.encode()).hexdigest()[:12]
return f"{wallet_prefix}-{digest}"
class _TrackerHTTPServer(http.server.HTTPServer):
def __init__(
self,
@@ -404,7 +446,10 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
self.send_header("Content-Type", "application/json")
self.send_header("Content-Length", str(len(body)))
self.end_headers()
self.wfile.write(body)
try:
self.wfile.write(body)
except BrokenPipeError:
pass
def _read_json_body(self) -> dict | None:
length = int(self.headers.get("Content-Length", 0))
@@ -484,6 +529,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
if not node.wallet_address or not server.contracts.registry.get_wallet(node.wallet_address).banned
]
data = []
seen_ids: set[str] = set()
for name, preset in server.model_presets.items():
model_nodes = [node for node in alive if node.model == name]
if not model_nodes:
@@ -494,25 +540,73 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
required_start,
required_end,
)
aliases = [name]
hf_repo = preset.get("hf_repo")
if hf_repo and hf_repo not in aliases:
aliases.append(hf_repo)
data.append({
"id": name,
"object": "model",
"created": created,
"owned_by": "meshnet",
"name": name,
"hf_repo": hf_repo,
"aliases": aliases,
"shard_coverage_percentage": coverage,
})
seen_ids.add(name)
hf_model_ids = sorted({
node.hf_repo or node.model
for node in alive
if node.model is not None
and node.model not in server.model_presets
and node.shard_start is not None
and node.shard_end is not None
and node.num_layers is not None
})
for model_id in hf_model_ids:
if model_id is None or model_id in seen_ids:
continue
model_nodes = [
node for node in alive
if node.shard_start is not None
and node.shard_end is not None
and node.num_layers is not None
and (node.hf_repo == model_id or (node.hf_repo is None and node.model == model_id))
]
if not model_nodes:
continue
short_names = sorted({node.model for node in model_nodes if node.model})
aliases = [model_id, *[name for name in short_names if name != model_id]]
required_start = 0
required_end = max(node.num_layers for node in model_nodes) - 1
data.append({
"id": model_id,
"object": "model",
"created": created,
"owned_by": "meshnet",
"name": short_names[0] if short_names else model_id,
"hf_repo": model_id if any(node.hf_repo == model_id for node in model_nodes) else None,
"aliases": aliases,
"shard_coverage_percentage": _coverage_percentage(
model_nodes,
required_start,
required_end,
),
})
seen_ids.add(model_id)
self._send_json(200, {"object": "list", "data": data})
def _handle_coverage(self, model: str):
server: _TrackerHTTPServer = self.server # type: ignore[assignment]
preset = server.model_presets.get(model)
if preset is None:
self._send_json(404, {"error": f"unknown model preset: {model!r}"})
return
required_start, required_end = _preset_layer_bounds(preset)
with server.lock:
self._purge_expired_nodes()
alive = [node for node in server.registry.values() if node.model == model]
resolved = _nodes_and_bounds_for_model(server, model)
if resolved is None:
self._send_json(404, {"error": f"no nodes registered for model {model!r}"})
return
alive, required_start, required_end = resolved
if server.contracts is not None:
alive = [
node for node in alive
@@ -821,7 +915,14 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
self._send_json(400, {"error": "num_layers must be an integer"})
return
node_id = str(uuid.uuid4())
node_id = _node_id_for_registration(
endpoint,
model,
wallet_address,
shard_start,
shard_end,
hf_repo,
)
entry = _NodeEntry(
node_id=node_id,
endpoint=endpoint.rstrip("/"),
@@ -1191,6 +1292,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
"route": [e.endpoint for e in route],
"nodes": [
{
"node_id": e.node_id,
"endpoint": e.endpoint,
"wallet_address": e.wallet_address,
"shard_start": e.shard_start,
@@ -1250,6 +1352,7 @@ class _TrackerHandler(http.server.BaseHTTPRequestHandler):
"route": [e.endpoint for e in route],
"nodes": [
{
"node_id": e.node_id,
"endpoint": e.endpoint,
"wallet_address": e.wallet_address,
"shard_start": e.shard_start,

View File

@@ -370,6 +370,11 @@ def test_real_model_startup_summary_shows_total_layers(tmp_path, monkeypatch, ca
lambda: {"device": "cpu", "gpu_name": None, "vram_mb": 0},
)
monkeypatch.setattr(startup_mod, "TorchNodeServer", FakeTorchNodeServer)
monkeypatch.setattr(
startup_mod,
"_post_json",
lambda _url, _payload, timeout=10.0: {"node_id": "node-test-123"},
)
node = run_startup(
tracker_url="http://127.0.0.1:8080",
@@ -380,8 +385,10 @@ def test_real_model_startup_summary_shows_total_layers(tmp_path, monkeypatch, ca
)
assert node.backend.total_layers == 24
assert node.tracker_node_id == "node-test-123"
output = capsys.readouterr().out
assert "Shard: layers 023; 24 of 24" in output
assert "Node ID: node-test-123" in output
# ---------------------------------------------------------------------------

View File

@@ -27,6 +27,69 @@ def _get_json(url: str) -> dict:
return json.loads(r.read())
def test_tracker_send_json_ignores_broken_pipe_after_client_disconnect():
"""A disconnected client must not dump a BrokenPipe traceback from the tracker."""
from meshnet_tracker.server import _TrackerHandler
class BrokenPipeWriter:
def write(self, _body):
raise BrokenPipeError
class DummyHandler:
wfile = BrokenPipeWriter()
def send_response(self, _status):
pass
def send_header(self, _name, _value):
pass
def end_headers(self):
pass
_TrackerHandler._send_json(DummyHandler(), 200, {"ok": True})
def test_tracker_registration_node_id_includes_wallet_prefix_and_stable_suffix():
tracker = TrackerServer()
tracker_port = tracker.start()
wallet = "7j77FsPY1evV8tuf4Z73AVrWwxBEW1pvKwi4EvcRD3g"
try:
first = _post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9100", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 0, "shard_end": 21, "wallet_address": wallet,
"hardware_profile": {}, "score": 1.0},
)
second = _post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9100", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 0, "shard_end": 21, "wallet_address": wallet,
"hardware_profile": {}, "score": 1.0},
)
different_endpoint = _post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9101", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 20, "shard_end": 23, "wallet_address": wallet,
"hardware_profile": {}, "score": 1.0},
)
assert first["node_id"].startswith("7j77FsPY-")
assert second["node_id"] == first["node_id"]
assert different_endpoint["node_id"].startswith("7j77FsPY-")
assert different_endpoint["node_id"] != first["node_id"]
route_resp = _get_json(
f"http://127.0.0.1:{tracker_port}/v1/route?model=Qwen/Qwen2.5-0.5B-Instruct"
)
assert route_resp["nodes"][0]["node_id"] == first["node_id"]
finally:
tracker.stop()
def test_tracker_node_registration():
"""A node can register with the tracker and receives a node_id."""
tracker = TrackerServer()
@@ -131,6 +194,64 @@ def test_tracker_coverage_endpoint_reports_uncovered_ranges():
tracker.stop()
def test_tracker_coverage_endpoint_accepts_registered_hf_repo_or_short_name():
"""Coverage endpoint supports real HF models registered outside preset catalog."""
tracker = TrackerServer()
tracker_port = tracker.start()
try:
_post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9101", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 0, "shard_end": 21, "hardware_profile": {}, "score": 1.0},
)
_post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9102", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 20, "shard_end": 23, "hardware_profile": {}, "score": 1.0},
)
by_repo = _get_json(
f"http://127.0.0.1:{tracker_port}/v1/coverage/Qwen/Qwen2.5-0.5B-Instruct"
)
by_short_name = _get_json(
f"http://127.0.0.1:{tracker_port}/v1/coverage/Qwen2.5-0.5B-Instruct"
)
assert by_repo["model"] == "Qwen/Qwen2.5-0.5B-Instruct"
assert by_repo["coverage"] == [
{"start_layer": 0, "end_layer": 19, "node_count": 1},
{"start_layer": 20, "end_layer": 21, "node_count": 2},
{"start_layer": 22, "end_layer": 23, "node_count": 1},
]
assert by_short_name["coverage"] == by_repo["coverage"]
finally:
tracker.stop()
def test_tracker_models_endpoint_lists_registered_hf_repo_and_short_name_alias():
tracker = TrackerServer()
tracker_port = tracker.start()
try:
_post_json(
f"http://127.0.0.1:{tracker_port}/v1/nodes/register",
{"endpoint": "http://127.0.0.1:9111", "model": "Qwen2.5-0.5B-Instruct",
"hf_repo": "Qwen/Qwen2.5-0.5B-Instruct", "num_layers": 24,
"shard_start": 0, "shard_end": 23, "hardware_profile": {}, "score": 1.0},
)
models_resp = _get_json(f"http://127.0.0.1:{tracker_port}/v1/models")
model = next(item for item in models_resp["data"] if item["id"] == "Qwen/Qwen2.5-0.5B-Instruct")
assert model["name"] == "Qwen2.5-0.5B-Instruct"
assert model["hf_repo"] == "Qwen/Qwen2.5-0.5B-Instruct"
assert model["aliases"] == ["Qwen/Qwen2.5-0.5B-Instruct", "Qwen2.5-0.5B-Instruct"]
assert model["shard_coverage_percentage"] == 100.0
finally:
tracker.stop()
def test_tracker_auto_assigns_new_node_to_uncovered_range_first():
"""Capability-driven registration fills the first uncovered layer gap."""
tracker = TrackerServer(model_presets={