emozilla
43e67d872f
Run models locally as a first-class provider. The CLI grows a managed llama.cpp runtime (engine install, model download, server supervision); the desktop app grows the full setup and management story on top of it. GUI surfaces ship behind the desktop --local launch flag (hermes desktop --local, or the flag on the packaged app); backend routes and the CLI are always live. Runtime (hermes_cli/local_runtime/): - curated GGUF catalog with per-machine variant selection: hardware probe (VRAM/RAM/UMA), fit planning with spill accounting, quant choice by context window - derived recommendation: quality-ranked picks gated by a predicted decode-speed floor, bandwidth-aware on unified memory; the decision table is pinned as a test (pick AND reason per memory class), and the Recommended badge explains its pick in a tooltip fed by the resolver's actual branch - engine install + model download with resumable split parts, cumulative plan-level progress, and staged-model integrity (a split GGUF counts only when every part is present) - server supervision: spawn/adopt/stop, router mode with per-model load progress relayed over SSE, abandoned-request cleanup Desktop: - Settings -> Providers -> Local models: one-click quickstart (install engine, download the recommended model, boot) plus per-model download/ activate/eject, fit-ranked catalog with context pills - model pickers (composer dropdown + Cmd+K) show staged local models, in-flight downloads as live progress rows, and load-into-memory bars - local-setup campaign tip for eligible hardware; System resources statusbar widget (GPU/VRAM/RAM); in-chat load progress during sends - friendly dead-server errors, and failed agent builds retry on the next send instead of wedging the session Co-developed with NVIDIA field feedback on RTX 5090 and DGX Spark.
115 lines
4.0 KiB
Python
115 lines
4.0 KiB
Python
"""Server on/off lifecycle route (round-9 feedback: 'we should be able to
|
|
completely turn off the local engine'). Contract: stop tears the server
|
|
down AND persists enabled=false (durable, unlike eject); start persists
|
|
enabled=true and boots."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
from fastapi.testclient import TestClient
|
|
|
|
|
|
@pytest.fixture
|
|
def client(tmp_path, monkeypatch):
|
|
monkeypatch.setenv("HERMES_HOME", str(tmp_path / ".hermes"))
|
|
from hermes_cli import web_server
|
|
|
|
test_client = TestClient(web_server.app)
|
|
token = getattr(web_server, "_SESSION_TOKEN", "")
|
|
if token:
|
|
test_client.headers["Authorization"] = f"Bearer {token}"
|
|
return test_client
|
|
|
|
|
|
def test_stop_disables_and_tears_down(client, monkeypatch):
|
|
stopped = {"called": False}
|
|
|
|
def _shutdown():
|
|
stopped["called"] = True
|
|
|
|
class _FakeSup:
|
|
pass
|
|
|
|
monkeypatch.setattr("hermes_cli.local_runtime.bootstrap.get_supervisor",
|
|
lambda: _FakeSup())
|
|
monkeypatch.setattr("hermes_cli.local_runtime.bootstrap.shutdown_local_runtime",
|
|
_shutdown)
|
|
|
|
r = client.post("/api/local-models/server", json={"action": "stop"})
|
|
assert r.status_code == 200
|
|
assert stopped["called"] is True
|
|
|
|
from hermes_cli.config import load_config
|
|
|
|
assert load_config()["local_runtime"]["enabled"] is False
|
|
|
|
|
|
def test_start_enables_and_boots(client, monkeypatch):
|
|
booted = {"called": False}
|
|
|
|
class _FakeSup:
|
|
base_url = "http://127.0.0.1:18434/v1"
|
|
|
|
def _ensure(config, force=False):
|
|
booted["called"] = True
|
|
assert force is True
|
|
return _FakeSup()
|
|
|
|
monkeypatch.setattr("hermes_cli.local_runtime.bootstrap.ensure_local_runtime",
|
|
_ensure)
|
|
|
|
r = client.post("/api/local-models/server", json={"action": "start"})
|
|
assert r.status_code == 200
|
|
assert booted["called"] is True
|
|
|
|
from hermes_cli.config import load_config
|
|
|
|
assert load_config()["local_runtime"]["enabled"] is True
|
|
|
|
|
|
def test_bogus_action_rejected(client):
|
|
r = client.post("/api/local-models/server", json={"action": "reboot"})
|
|
assert r.status_code == 400
|
|
|
|
|
|
def test_status_reports_loaded_models_from_live_router(client, monkeypatch):
|
|
"""Round-11 regression: the loaded-models read inside the status route
|
|
raised NameError (missing json import), the blanket except swallowed it,
|
|
and {} shipped as truth — 'Not in memory' on a machine with 30 GB of
|
|
VRAM in use. This test exercises the REAL route against a stub router
|
|
and demands the loaded set comes through."""
|
|
import http.server
|
|
import json as _json
|
|
import threading
|
|
|
|
class _Router(http.server.BaseHTTPRequestHandler):
|
|
def do_GET(self):
|
|
body = _json.dumps({"data": [
|
|
{"id": "m-loaded", "status": {"value": "loaded"}},
|
|
{"id": "m-loading", "status": {"value": "loading"}},
|
|
{"id": "m-cold", "status": {"value": "unloaded"}},
|
|
]}).encode()
|
|
self.send_response(200)
|
|
self.send_header("Content-Length", str(len(body)))
|
|
self.end_headers()
|
|
self.wfile.write(body)
|
|
|
|
def log_message(self, *a):
|
|
pass
|
|
|
|
server = http.server.HTTPServer(("127.0.0.1", 0), _Router)
|
|
threading.Thread(target=server.serve_forever, daemon=True).start()
|
|
try:
|
|
port = server.server_address[1]
|
|
# Patch the ROUTE's binding: local_models binds _state_endpoint via
|
|
# from-import at module load, so patching the endpoint module's
|
|
# attribute never reaches the name the route actually calls.
|
|
monkeypatch.setattr(
|
|
"hermes_cli.web_routers.local_models._state_endpoint",
|
|
lambda: {"base_url": f"http://127.0.0.1:{port}/v1", "api_key": "k"})
|
|
payload = client.get("/api/local-models/status").json()
|
|
assert payload["server_running"] is True
|
|
assert payload["loaded_models"] == {"m-loaded": "loaded", "m-loading": "loading"}
|
|
finally:
|
|
server.shutdown()
|