Teknium
39975613b1
Second, deeper pass over tools/gateway/hermes_cli plus first pass over the trees wave 1 missed (acp, acp_adapter, skills, computer_use, docker, dashboard, conformance, monitoring, secret_sources, hermes_state, providers). Same rubric as wave 1 (AGENTS.md test policy); security, alternation/caching invariants, issue-number regressions, and E2E kept. Real test-quality fixes found and rooted out along the way: - tests/tools/test_command_guards.py made real auxiliary-LLM HTTPS calls (DEFAULT_CONFIG smart-approval leaked in) — pinned approval mode=manual via autouse fixture: 17.4s → 0.4s. - test_model_switch_custom_providers.py / test_user_providers_model_switch.py silently probed live provider catalogs (~2s/test) — stubbed cached_provider_model_ids/provider_model_ids/fetch_api_models. - test_telegram_noise_filter.py: 15-platform copy-paste matrix over shared gateway.run logic → 3 representative platforms (55s → 3.9s). - test_gateway_shutdown.py: stop()'s 5s interrupt-deadline loop spun on MagicMock agents — interrupt.side_effect now clears _running_agents (22s → 1.0s). - test_gateway_inactivity_timeout.py poll-harness timings shrunk 3-5x (24s → 1.1s); test_mcp_stability.py backoff/SIGTERM-grace sleeps patched (15.4s → 2.5s); test_async_delegation.py negative-drain wait 5s → 0.5s. - test_telegram_init_deadline.py: loop-block margin restored to 1.0s with rationale comment — the watchdog-dump assertion needs the loop blocked well past deadline+grace under parallel load (flaked once in the 40-worker verification run at a 0.2s margin). Verification: full hermetic suite via scripts/run_tests.sh — 2,438 files, 21,718 tests passed, 0 failed, 293.9s wall. Suite totals vs original baseline: 46,820 → 19,757 test functions (−57.8%), wall 583.5s → 293.9s (−50%), subprocess CPU 13,564s → 11,623s.
116 lines
4.0 KiB
Python
116 lines
4.0 KiB
Python
"""Regression tests for issue #66092.
|
|
|
|
Streamable-HTTP / SSE MCP transports run their stream pump inside an anyio
|
|
TaskGroup. A transient stream drop (idle timeout, brief backend blip) surfaces
|
|
as a ``BaseExceptionGroup`` escaping the transport context manager. Before the
|
|
fix that group reached ``run()``'s error path, which applied exponential
|
|
backoff and eventually parked the server for 300s and deregistered its tools —
|
|
a multi-minute tool outage for a sub-second glitch.
|
|
|
|
``_run_http`` now converts such a transport TaskGroup failure into a clean
|
|
``"reconnect"`` (immediate rebuild, no backoff/park), while still propagating
|
|
real shutdown/cancellation and genuine connect/handshake failures.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import asyncio
|
|
|
|
import pytest
|
|
|
|
from tools.mcp_tool import MCPServerTask
|
|
|
|
|
|
def _group(*excs) -> BaseExceptionGroup:
|
|
return BaseExceptionGroup("transport drop", list(excs))
|
|
|
|
|
|
# ── Unit coverage for the decision helper ────────────────────────────────────
|
|
|
|
class TestReconnectOrReraiseGroup:
|
|
def test_live_session_transient_drop_reconnects(self):
|
|
task = MCPServerTask("t")
|
|
task._ready.set() # a live session was established this attempt
|
|
assert task._reconnect_or_reraise_group(
|
|
_group(ConnectionError("sse stream dropped"))
|
|
) == "reconnect"
|
|
|
|
|
|
def test_no_live_session_reraises_for_backoff(self):
|
|
task = MCPServerTask("t")
|
|
# _ready NOT set: a connect/handshake failure, not a transient drop —
|
|
# let run()'s backoff handle it instead of hot-looping.
|
|
with pytest.raises(BaseExceptionGroup):
|
|
task._reconnect_or_reraise_group(_group(ConnectionError("handshake")))
|
|
|
|
|
|
# ── Integration coverage: real _run_http new-HTTP branch ─────────────────────
|
|
|
|
class _RaisingTransportCM:
|
|
"""Fake streamable-HTTP transport whose __aexit__ raises a group, mimicking
|
|
the SDK's anyio TaskGroup tearing down on a stream drop."""
|
|
|
|
def __init__(self, exc):
|
|
self._exc = exc
|
|
|
|
async def __aenter__(self):
|
|
return (object(), object(), lambda: "session-id")
|
|
|
|
async def __aexit__(self, *exc_info):
|
|
raise self._exc
|
|
|
|
|
|
class _FakeSession:
|
|
async def initialize(self):
|
|
return object()
|
|
|
|
|
|
class _FakeSessionCM:
|
|
async def __aenter__(self):
|
|
return _FakeSession()
|
|
|
|
async def __aexit__(self, *exc_info):
|
|
return False
|
|
|
|
|
|
def _make_http_task(monkeypatch, transport_exc):
|
|
task = MCPServerTask("http-drop")
|
|
|
|
monkeypatch.setattr("tools.mcp_tool._MCP_HTTP_AVAILABLE", True, raising=False)
|
|
monkeypatch.setattr("tools.mcp_tool._MCP_NEW_HTTP", True, raising=False)
|
|
monkeypatch.setattr(
|
|
"tools.mcp_tool.streamable_http_client",
|
|
lambda url, http_client=None: _RaisingTransportCM(transport_exc),
|
|
raising=False,
|
|
)
|
|
monkeypatch.setattr(
|
|
"tools.mcp_tool.ClientSession",
|
|
lambda *a, **k: _FakeSessionCM(),
|
|
raising=False,
|
|
)
|
|
|
|
async def _no_discover(self):
|
|
return None
|
|
|
|
async def _reconnect_lifecycle(self):
|
|
return "reconnect"
|
|
|
|
monkeypatch.setattr(MCPServerTask, "_discover_tools", _no_discover)
|
|
monkeypatch.setattr(MCPServerTask, "_wait_for_lifecycle_event", _reconnect_lifecycle)
|
|
return task
|
|
|
|
|
|
def test_run_http_transport_group_returns_reconnect(monkeypatch):
|
|
task = _make_http_task(monkeypatch, _group(ConnectionError("stream dropped")))
|
|
reason = asyncio.run(task._run_http({"url": "http://127.0.0.1:9/mcp"}))
|
|
assert reason == "reconnect"
|
|
# We had a live session; readiness stays set for run() to clear on re-entry.
|
|
assert task._ready.is_set()
|
|
|
|
|
|
def test_run_http_transport_group_reraises_on_shutdown(monkeypatch):
|
|
task = _make_http_task(monkeypatch, _group(ConnectionError("stream dropped")))
|
|
task._shutdown_event.set()
|
|
with pytest.raises(BaseExceptionGroup):
|
|
asyncio.run(task._run_http({"url": "http://127.0.0.1:9/mcp"}))
|