feat(tenancy): add Workspace multi-tenant foundation (#2353)

* Document multi-tenant workspace architecture

* Add OSS and commercial workspace boundaries

* docs: redesign multi-tenant workspace architecture

* feat(tenancy): implement workspace isolation

* docs(tenancy): record verification evidence

* docs(tenancy): revise single-instance SaaS topology

* docs(tenancy): refine architecture options

* docs: finalize cloud v2 multi-tenant decisions

* feat(tenancy): establish cloud isolation foundations

* feat(tenancy): harden shared cloud runtime boundaries

* docs(tenancy): record final isolation verification

* fix(tenancy): close isolation and permission gaps

* docs(tenancy): record final isolation verification

* feat(tenancy): connect cloud workspace control plane

* fix(build): install git for pinned SDK

* docs(cloud): update control plane verification

* chore: update multi-tenant SDK pin

* fix(cloud): skip legacy model sync during startup

* test(cloud): preserve minimal model manager fixtures

* fix(cloud): preserve authenticated account context

* fix(cloud): reuse authenticated account for user info

* feat(cloud): complete Workspace settings navigation

* test(web): cover Workspace dropdown menu

* feat(web): place workspace controls in sidebar

* refactor(web): streamline workspace controls

* style(web): format workspace layout test

* fix(cloud): surface runtime and workspace plan status

* fix(plugin): keep runtime identity stable across restarts

* fix(ui): widen and center workspace switcher

* fix(ui): hide roles from workspace switcher

* fix(ui): align workspace switcher with sidebar entries

* feat(workspace): add in-product collaboration and direct Cloud launch

* style: format collaboration changes

* fix(workspace): bind collaboration APIs to tenant UoW

* fix(cloud): preserve Core-owned collaboration state

* test(cloud): require Space identity for invite registration

* feat(cloud): complete secure invitation experience

* style(web): format invitation flows

* fix(cloud): recover box runtime without unscoped skill reload

* feat(oss): enforce invitation account and owner billing flows

* style: format OSS account service

* test(oss): cover invitation logout handoff

* fix(oss): resolve workspace owner in scoped session

* feat(cloud): harden multi-tenant runtime resources

* fix(cloud): bound runtime restart storms

* fix(cloud): eliminate periodic runtime CPU spikes

* fix(cloud): enforce instance capacity ceilings

* fix(cloud): scope public login capability discovery

* fix(cloud): bound tenant maintenance and monitoring work

* fix(runtime): bound tenant resource amplification

* fix(deps): pin green multi-tenant plugin SDK

* fix(cloud): handle unavailable skill capability

* fix(security): require authentication for image file endpoint (H-2)

- Changed /api/v1/files/image from AuthType.NONE to USER_TOKEN_OR_API_KEY
- Added Permission.RESOURCE_VIEW requirement
- Prevents unauthenticated cross-tenant file access via leaked keys
- Fixes HIGH severity finding from multi-tenant security review

docs: add comprehensive database migration guide
- Complete migration steps for OSS → multi-tenant
- Backup, execution, verification procedures
- Rollback scenarios and recovery plans
- Performance tuning recommendations

* test: add comprehensive cross-tenant isolation tests

Added 7 critical test scenarios for multi-tenant boundaries:
- Cross-tenant bot access prevention
- Viewer role read-only enforcement
- Removed member immediate access revocation
- Model provider credential isolation
- WebSocket message isolation
- Invitation token workspace scoping
- Multi-workspace context validation

These tests address P0-2 coverage gaps for:
- workspaces.py (membership & invitation flows)
- user.py (authentication & authorization)
- websocket_chat.py (real-time isolation)
- plugins.py (resource access control)

docs: finalize database migration guide

* fix(security): resolve M-1, M-2, M-3 security findings

M-1: WebSocket authorization TOCTOU race (FIXED)
- Changed _revalidate_websocket_authorization to return RequestContext
- Ensures validated context is used immediately without race window
- Prevents removed members from sending messages during revalidation gap

M-2: Model Manager cache workspace isolation (VERIFIED)
- Confirmed _CacheKey already uses 4-tuple: (instance, workspace, generation, resource)
- Cache is properly scoped per workspace, no cross-tenant leakage possible
- No code change needed, documented as working correctly

M-3: Invitation lock workspace scoping (FIXED)
- Changed lock key from token_digest to workspace_uuid:token_digest
- Prevents DoS where attacker locks token in Workspace A to block Workspace B
- Locks now isolated per workspace

All MEDIUM severity findings from security review now resolved.

* fix(cloud): unblock tenant CI and enforce knowledge quotas

* fix(tenancy): scope rerank model sync

---------

Co-authored-by: dadachann <185672915+dadachann@users.noreply.github.com>
This commit is contained in:
RockChinQ
2026-07-30 21:43:35 +08:00
committed by GitHub
parent 463b120923
commit e1ac5e0fc8
468 changed files with 78320 additions and 13137 deletions
@@ -0,0 +1,278 @@
from __future__ import annotations
import asyncio
import threading
import pytest
from langbot.pkg.utils.bounded_executor import (
BlockingWorkCapacityError,
BoundedThreadPoolExecutor,
blocking_work_scope,
configure_bounded_default_executor,
run_blocking_atomic,
run_blocking_cleanup,
)
def test_bounded_executor_rejects_instead_of_queueing_without_limit():
executor = BoundedThreadPoolExecutor(
max_workers=1,
max_pending=1,
max_inflight_per_scope=1,
)
started = threading.Event()
release = threading.Event()
def block() -> str:
started.set()
release.wait(timeout=5)
return 'done'
first = executor.submit(block)
assert started.wait(timeout=1)
second = executor.submit(lambda: 'queued')
with pytest.raises(
BlockingWorkCapacityError,
match='capacity reached',
):
executor.submit(lambda: 'rejected')
assert executor.snapshot() == {
'max_workers': 1,
'max_pending': 1,
'max_inflight_per_scope': 1,
'inflight': 2,
'running': 1,
'pending': 1,
'active_scopes': 0,
'submitted_total': 2,
'completed_total': 0,
'rejected_total': 1,
'global_rejected_total': 1,
'scope_rejected_total': 0,
}
release.set()
assert first.result(timeout=1) == 'done'
assert second.result(timeout=1) == 'queued'
assert executor.snapshot()['inflight'] == 0
executor.shutdown()
def test_workspace_scope_cannot_monopolize_global_workers():
executor = BoundedThreadPoolExecutor(
max_workers=2,
max_pending=2,
max_inflight_per_scope=1,
)
release = threading.Event()
workspace_a_started = threading.Event()
workspace_b_started = threading.Event()
def block(started: threading.Event) -> str:
started.set()
release.wait(timeout=5)
return 'done'
try:
with blocking_work_scope('workspace-a'):
workspace_a = executor.submit(block, workspace_a_started)
assert workspace_a_started.wait(timeout=1)
with pytest.raises(
BlockingWorkCapacityError,
match='Workspace blocking executor capacity reached',
):
executor.submit(lambda: 'rejected')
with blocking_work_scope('workspace-b'):
workspace_b = executor.submit(block, workspace_b_started)
assert workspace_b_started.wait(timeout=1)
snapshot = executor.snapshot()
assert snapshot['inflight'] == 2
assert snapshot['active_scopes'] == 2
assert snapshot['scope_rejected_total'] == 1
assert snapshot['global_rejected_total'] == 0
finally:
release.set()
assert workspace_a.result(timeout=1) == 'done'
assert workspace_b.result(timeout=1) == 'done'
executor.shutdown()
def test_default_executor_bounds_asyncio_to_thread():
loop = asyncio.new_event_loop()
executor = configure_bounded_default_executor(
loop,
max_workers=2,
max_pending=3,
)
try:
assert loop.run_until_complete(asyncio.to_thread(lambda: 'bounded')) == 'bounded'
assert executor.snapshot()['completed_total'] == 1
assert (
configure_bounded_default_executor(
loop,
max_workers=2,
max_pending=3,
)
is executor
)
finally:
executor.shutdown()
loop.close()
def test_workspace_scope_is_enforced_for_asyncio_to_thread():
loop = asyncio.new_event_loop()
executor = configure_bounded_default_executor(
loop,
max_workers=2,
max_pending=2,
max_inflight_per_scope=1,
)
started = threading.Event()
release = threading.Event()
def block() -> str:
started.set()
release.wait(timeout=5)
return 'workspace-a'
async def exercise() -> None:
with blocking_work_scope('workspace-a'):
workspace_a = asyncio.create_task(asyncio.to_thread(block))
try:
while not started.is_set():
await asyncio.sleep(0)
with blocking_work_scope('workspace-a'):
with pytest.raises(
BlockingWorkCapacityError,
match='Workspace blocking executor capacity reached',
):
await asyncio.to_thread(lambda: 'rejected')
with blocking_work_scope('workspace-b'):
assert await asyncio.to_thread(lambda: 'workspace-b') == 'workspace-b'
finally:
release.set()
assert await workspace_a == 'workspace-a'
try:
loop.run_until_complete(exercise())
finally:
executor.shutdown()
loop.close()
def test_blocking_cleanup_waits_for_capacity_instead_of_leaking_work():
loop = asyncio.new_event_loop()
executor = configure_bounded_default_executor(
loop,
max_workers=1,
max_pending=0,
max_inflight_per_scope=1,
)
started = threading.Event()
release = threading.Event()
cleaned = threading.Event()
def block() -> None:
started.set()
release.wait(timeout=5)
async def exercise() -> None:
blocker = asyncio.create_task(asyncio.to_thread(block))
while not started.is_set():
await asyncio.sleep(0)
cleanup = asyncio.create_task(run_blocking_cleanup(cleaned.set))
await asyncio.sleep(0.03)
assert not cleanup.done()
release.set()
await blocker
await cleanup
try:
loop.run_until_complete(exercise())
assert cleaned.is_set()
assert executor.snapshot()['global_rejected_total'] >= 1
finally:
release.set()
executor.shutdown()
loop.close()
def test_blocking_atomic_waits_for_thread_before_propagating_cancellation():
loop = asyncio.new_event_loop()
executor = configure_bounded_default_executor(
loop,
max_workers=1,
max_pending=1,
max_inflight_per_scope=1,
)
started = threading.Event()
release = threading.Event()
completed = threading.Event()
def block() -> None:
started.set()
release.wait(timeout=5)
completed.set()
async def exercise() -> None:
operation = asyncio.create_task(run_blocking_atomic(block))
while not started.is_set():
await asyncio.sleep(0)
operation.cancel()
await asyncio.sleep(0)
assert not operation.done()
release.set()
with pytest.raises(asyncio.CancelledError):
await operation
try:
loop.run_until_complete(exercise())
assert completed.is_set()
finally:
release.set()
executor.shutdown()
loop.close()
@pytest.mark.parametrize(
('max_workers', 'max_pending'),
[
(0, 1),
(65, 1),
(1, -1),
(1, 4097),
(True, 1),
],
)
def test_bounded_executor_rejects_unsafe_limits(
max_workers,
max_pending,
):
with pytest.raises(ValueError):
BoundedThreadPoolExecutor(
max_workers=max_workers,
max_pending=max_pending,
)
@pytest.mark.parametrize(
('max_workers', 'max_inflight_per_scope'),
[(8, 0), (8, 4097), (8, True), (8, 5), (2, 2)],
)
def test_bounded_executor_rejects_unsafe_scope_limits(
max_workers,
max_inflight_per_scope,
):
with pytest.raises(ValueError):
BoundedThreadPoolExecutor(
max_workers=max_workers,
max_inflight_per_scope=max_inflight_per_scope,
)
@@ -0,0 +1,577 @@
from __future__ import annotations
import io
import json
from collections import deque
from pathlib import Path
import pytest
from scripts import cloud_runtime_soak as soak
def _write(path: Path, value: str) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
path.write_text(value, encoding='utf-8')
def _process_stat(pid: int, *, user_ticks: int, system_ticks: int) -> str:
fields = [
'S',
'0',
'0',
'0',
'0',
'0',
'0',
'0',
'0',
'0',
'0',
str(user_ticks),
str(system_ticks),
'0',
'0',
'0',
'0',
'0',
'0',
'0',
]
return f'{pid} (worker with spaces) ' + ' '.join(fields)
def _sample(timestamp: float, **metrics: float) -> soak.MetricSample:
return soak.MetricSample(
monotonic_seconds=timestamp,
wall_time=f'sample-{timestamp}',
metrics=metrics,
)
def _state(
kind: str,
samples: list[soak.MetricSample],
*,
baseline: dict[str, float] | None = None,
latest: dict[str, float] | None = None,
) -> soak.TargetState:
return soak.TargetState(
target=soak.Target(name='target', kind=kind, location='/target'),
samples=deque(samples),
baseline_metrics=baseline or dict(samples[0].metrics),
last_metrics=latest or dict(samples[-1].metrics),
attempted_samples=len(samples),
successful_samples=len(samples),
)
def _thresholds(**overrides) -> soak.Thresholds:
values = {
'max_memory_growth_bytes': 64 * soak.BYTES_PER_MIB,
'max_memory_slope_bytes_per_hour': 32 * soak.BYTES_PER_MIB,
'max_tail_cpu_cores': 0.5,
'max_throttled_period_ratio': 0.25,
'allow_rejections': False,
'max_transient_gauge_growth': 0,
'require_hard_limits': False,
'max_event_loop_lag_ms': 1000,
'max_event_loop_p95_lag_ms': 250,
'require_event_loop_metrics': True,
}
values.update(overrides)
return soak.Thresholds(**values)
@pytest.mark.parametrize(
('value', 'expected'),
[
('1', 1),
('1.5s', 1.5),
('2m', 120),
('3H', 10_800),
('1d', 86_400),
],
)
def test_parse_duration(value: str, expected: float) -> None:
assert soak.parse_duration(value) == expected
@pytest.mark.parametrize('value', ['', '0s', '-1s', 'wat'])
def test_parse_duration_rejects_invalid_values(value: str) -> None:
with pytest.raises(Exception):
soak.parse_duration(value)
def test_build_targets_rejects_secret_bearing_health_url() -> None:
with pytest.raises(ValueError, match='must not contain credentials'):
soak.build_targets(
endpoints=['core=https://user:secret@example.test/healthz'],
cgroups=[],
pids=[],
)
with pytest.raises(ValueError, match='query parameters'):
soak.build_targets(
endpoints=['core=https://example.test/healthz?token=secret'],
cgroups=[],
pids=[],
)
def test_read_cgroup_snapshot_reads_v2_pressure_and_limits(tmp_path: Path) -> None:
_write(tmp_path / 'memory.current', '1048576\n')
_write(tmp_path / 'memory.peak', '2097152\n')
_write(tmp_path / 'memory.swap.current', '4096\n')
_write(tmp_path / 'memory.max', '1073741824\n')
_write(tmp_path / 'memory.swap.max', 'max\n')
_write(tmp_path / 'pids.current', '7\n')
_write(tmp_path / 'pids.max', '128\n')
_write(
tmp_path / 'cpu.stat',
'usage_usec 1234\nnr_periods 20\nnr_throttled 2\nthrottled_usec 99\n',
)
_write(tmp_path / 'memory.events', 'high 1\nmax 2\noom 0\noom_kill 0\n')
_write(tmp_path / 'pids.events', 'max 3\n')
_write(tmp_path / 'cpu.max', '100000 100000\n')
metrics = soak.read_cgroup_snapshot(tmp_path)
assert metrics['memory.current_bytes'] == 1_048_576
assert metrics['memory.max_bytes'] == 1_073_741_824
assert 'memory.swap.max_bytes' not in metrics
assert metrics['cpu.usage_usec'] == 1234
assert metrics['cpu.nr_throttled'] == 2
assert metrics['memory.events.max'] == 2
assert metrics['pids.events.max'] == 3
assert metrics['cpu.quota_usec'] == 100_000
def test_read_process_snapshot_aggregates_descendants(tmp_path: Path) -> None:
for pid, rss_kib, threads, user_ticks, system_ticks in (
(100, 1000, 2, 100, 50),
(200, 500, 1, 20, 10),
):
process_root = tmp_path / str(pid)
_write(
process_root / 'status',
f'Name:\tworker\nVmRSS:\t{rss_kib} kB\nThreads:\t{threads}\n',
)
_write(
process_root / 'stat',
_process_stat(
pid,
user_ticks=user_ticks,
system_ticks=system_ticks,
),
)
(process_root / 'fd').mkdir()
(process_root / 'fd' / '0').touch()
(process_root / 'fd' / '1').touch()
(process_root / 'task' / str(pid)).mkdir(parents=True)
_write(tmp_path / '100' / 'task' / '100' / 'children', '200\n')
_write(tmp_path / '200' / 'task' / '200' / 'children', '\n')
metrics = soak.read_process_snapshot(100, proc_root=tmp_path, clock_ticks=100)
assert metrics == {
'rss_bytes': 1500 * 1024,
'cpu_seconds': 1.8,
'threads': 3,
'open_fds': 4,
'processes': 2,
}
class _FakeResponse:
def __init__(self, payload: dict) -> None:
self.status = 200
self.headers = {'Content-Type': 'application/json'}
self._body = json.dumps(payload).encode()
def __enter__(self):
return self
def __exit__(self, *_args):
return None
def getcode(self) -> int:
return self.status
def read(self, limit: int) -> bytes:
return self._body[:limit]
def test_read_endpoint_snapshot_flattens_resource_metrics() -> None:
def opener(_request, *, timeout: float):
assert timeout == 2
return _FakeResponse(
{
'code': 0,
'resources': {
'blocking_executor': {
'pending': 0,
'global_rejected_total': 2,
},
'restart_coordinator': {
'active_launches': 1,
'circuit_open_total': 3,
},
},
}
)
metrics = soak.read_endpoint_snapshot(
'http://langbot.test/healthz',
timeout_seconds=2,
opener=opener,
)
assert metrics['http.ok'] == 1
assert metrics['body.resources.blocking_executor.pending'] == 0
assert metrics['body.resources.blocking_executor.global_rejected_total'] == 2
assert metrics['body.resources.restart_coordinator.active_launches'] == 1
assert metrics['body.resources.restart_coordinator.circuit_open_total'] == 3
def test_read_endpoint_snapshot_fails_closed_on_not_ready() -> None:
def opener(_request, *, timeout: float):
return _FakeResponse({'ready': False})
with pytest.raises(RuntimeError, match='not ready'):
soak.read_endpoint_snapshot(
'http://box.test/readyz',
timeout_seconds=2,
opener=opener,
)
def test_evaluate_gate_accepts_stable_process_tail() -> None:
state = _state(
'process',
[
_sample(0, rss_bytes=100 * soak.BYTES_PER_MIB, cpu_seconds=0),
_sample(1800, rss_bytes=101 * soak.BYTES_PER_MIB, cpu_seconds=10),
_sample(3600, rss_bytes=100 * soak.BYTES_PER_MIB, cpu_seconds=20),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert result.passed
assert result.targets['process:target']['cpu.average_cores'] < 0.01
def test_evaluate_gate_detects_material_memory_leak_and_idle_cpu() -> None:
state = _state(
'process',
[
_sample(0, rss_bytes=100 * soak.BYTES_PER_MIB, cpu_seconds=0),
_sample(1800, rss_bytes=150 * soak.BYTES_PER_MIB, cpu_seconds=1800),
_sample(3600, rss_bytes=200 * soak.BYTES_PER_MIB, cpu_seconds=3600),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert not result.passed
assert any('grew 100.00 MiB' in failure for failure in result.failures)
assert any('tail CPU averaged 1.000 cores' in failure for failure in result.failures)
def test_evaluate_gate_counts_oom_and_throttling_across_workload() -> None:
baseline = {
'memory.current_bytes': 100,
'memory.events.high': 0,
'memory.events.max': 0,
'memory.events.oom': 0,
'memory.events.oom_kill': 0,
'pids.events.max': 0,
'cpu.usage_usec': 0,
'cpu.nr_periods': 0,
'cpu.nr_throttled': 0,
}
latest = {
**baseline,
'memory.events.oom_kill': 1,
'pids.events.max': 2,
'cpu.usage_usec': 1_000_000,
'cpu.nr_periods': 100,
'cpu.nr_throttled': 30,
}
state = _state(
'cgroup',
[
_sample(100, **{**baseline, 'cpu.usage_usec': 500_000}),
_sample(200, **latest),
],
baseline=baseline,
latest=latest,
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=100,
thresholds=_thresholds(max_tail_cpu_cores=100),
)
assert any('memory.events.oom_kill by 1' in failure for failure in result.failures)
assert any('pids.events.max by 2' in failure for failure in result.failures)
assert any('throttled-period ratio 0.300' in failure for failure in result.failures)
def test_evaluate_gate_can_require_all_hard_cgroup_limits() -> None:
metrics = {
'memory.current_bytes': 100,
'memory.max_bytes': 1000,
'memory.events.high': 0,
'memory.events.max': 0,
'memory.events.oom': 0,
'memory.events.oom_kill': 0,
'pids.current': 1,
'pids.events.max': 0,
'cpu.usage_usec': 0,
'cpu.nr_periods': 0,
'cpu.nr_throttled': 0,
}
state = _state(
'cgroup',
[_sample(0, **metrics), _sample(60, **metrics)],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(require_hard_limits=True),
)
assert any('missing hard cgroup limits: cpu, pids, swap' in failure for failure in result.failures)
def test_evaluate_gate_detects_executor_rejection_and_stuck_pending() -> None:
prefix = 'body.resources.blocking_executor'
state = _state(
'endpoint',
[
_sample(
0,
**{
f'{prefix}.pending': 1,
f'{prefix}.global_rejected_total': 0,
},
),
_sample(
60,
**{
f'{prefix}.pending': 2,
f'{prefix}.global_rejected_total': 1,
},
),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('global_rejected_total by 1' in failure for failure in result.failures)
assert any('pending above zero' in failure for failure in result.failures)
def test_evaluate_gate_detects_restart_circuit_and_stuck_launch() -> None:
prefix = 'body.resources.restart_coordinator'
state = _state(
'endpoint',
[
_sample(
0,
**{
f'{prefix}.active_launches': 1,
f'{prefix}.gate_waiters': 2,
f'{prefix}.half_open_probe_inflight': 1,
f'{prefix}.open_remaining_seconds': 60,
f'{prefix}.circuit_open_total': 0,
},
),
_sample(
60,
**{
f'{prefix}.active_launches': 1,
f'{prefix}.gate_waiters': 2,
f'{prefix}.half_open_probe_inflight': 1,
f'{prefix}.open_remaining_seconds': 1,
f'{prefix}.circuit_open_total': 1,
},
),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('circuit_open_total by 1' in failure for failure in result.failures)
assert any('active_launches above zero' in failure for failure in result.failures)
assert any('gate_waiters above zero' in failure for failure in result.failures)
assert any('half_open_probe_inflight above zero' in failure for failure in result.failures)
assert any('open_remaining_seconds above zero' in failure for failure in result.failures)
def test_evaluate_gate_detects_stuck_mcp_projection_cleanup() -> None:
prefix = 'body.resources.runtimes'
state = _state(
'endpoint',
[
_sample(
0,
**{
f'{prefix}.mcp_projection_retirements': 3,
f'{prefix}.mcp_projection_reconcile_active': 1,
},
),
_sample(
60,
**{
f'{prefix}.mcp_projection_retirements': 1,
f'{prefix}.mcp_projection_reconcile_active': 1,
},
),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('mcp_projection_retirements above zero' in failure for failure in result.failures)
assert any('mcp_projection_reconcile_active above zero' in failure for failure in result.failures)
def test_evaluate_gate_detects_directory_and_database_capacity_violation() -> None:
prefix = 'body.resources'
state = _state(
'endpoint',
[
_sample(
0,
**{
f'{prefix}.directory.active_workspaces': 1000,
f'{prefix}.directory.max_active_workspaces': 1000,
f'{prefix}.database_pool.checked_out': 20,
f'{prefix}.database_pool.configured_capacity': 20,
},
),
_sample(
60,
**{
f'{prefix}.directory.active_workspaces': 1001,
f'{prefix}.directory.max_active_workspaces': 1000,
f'{prefix}.database_pool.checked_out': 21,
f'{prefix}.database_pool.configured_capacity': 20,
},
),
],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('directory.active_workspaces' in failure for failure in result.failures)
assert any('database_pool.checked_out' in failure for failure in result.failures)
def test_evaluate_gate_requires_capacity_gauges_from_named_core_endpoint() -> None:
state = _state(
'endpoint',
[_sample(0, **{'http.ok': 1}), _sample(60, **{'http.ok': 1})],
)
state.target = soak.Target(name='core', kind='endpoint', location='/core')
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(require_event_loop_metrics=False),
)
assert any('required capacity gauge' in failure for failure in result.failures)
def test_evaluate_gate_detects_event_loop_stall_and_sustained_lag() -> None:
prefix = 'body.resources.event_loop'
samples = [
_sample(
0,
**{
f'{prefix}.running': 1,
f'{prefix}.samples_total': 10,
f'{prefix}.recent_max_lag_ms': 20,
f'{prefix}.recent_p95_lag_ms': 10,
},
),
_sample(
60,
**{
f'{prefix}.running': 1,
f'{prefix}.samples_total': 70,
f'{prefix}.recent_max_lag_ms': 1500,
f'{prefix}.recent_p95_lag_ms': 300,
},
),
]
state = _state('endpoint', samples)
state.observed_max_metrics = {
metric: max(sample.metrics[metric] for sample in samples) for metric in samples[0].metrics
}
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('event-loop lag reached 1500.00 ms' in item for item in result.failures)
assert any('recent p95 reached 300.00 ms' in item for item in result.failures)
def test_evaluate_gate_requires_running_event_loop_monitor() -> None:
state = _state(
'endpoint',
[_sample(0, **{'http.ok': 1}), _sample(60, **{'http.ok': 1})],
)
result = soak.evaluate_gate(
[state],
analysis_start_seconds=0,
thresholds=_thresholds(),
)
assert any('did not expose event-loop health metrics' in item for item in result.failures)
def test_write_json_line_streams_one_record() -> None:
stream = io.StringIO()
soak._write_json_line(stream, {'z': 1, 'a': 2})
assert stream.getvalue() == '{"a":2,"z":1}\n'
def test_main_requires_a_target() -> None:
with pytest.raises(SystemExit) as exc_info:
soak.main(['--duration', '2s', '--sample-interval', '1s', '--startup-grace', '1s'])
assert exc_info.value.code == 2
@@ -0,0 +1,75 @@
from __future__ import annotations
import asyncio
import time
import pytest
from langbot.pkg.utils.event_loop_monitor import EventLoopLagMonitor
@pytest.mark.parametrize(
'kwargs',
[
{'sample_interval_seconds': 0},
{'sample_interval_seconds': float('inf')},
{'recent_sample_count': 1},
{'recent_sample_count': 3601},
],
)
def test_event_loop_monitor_rejects_unbounded_configuration(kwargs) -> None:
with pytest.raises(ValueError):
EventLoopLagMonitor(**kwargs)
def test_event_loop_monitor_snapshot_is_bounded_and_reports_p95() -> None:
monitor = EventLoopLagMonitor(recent_sample_count=4)
for lag_seconds in (0.001, 0.002, 0.003, 0.004, 0.100):
monitor._record_lag_seconds(lag_seconds)
snapshot = monitor.snapshot()
assert snapshot == {
'running': False,
'samples_total': 5,
'last_lag_ms': 100,
'recent_p95_lag_ms': 100,
'recent_max_lag_ms': 100,
'max_lag_ms': 100,
}
assert len(monitor._recent_lag_ms) == 4
async def test_event_loop_monitor_start_and_stop_are_idempotent() -> None:
monitor = EventLoopLagMonitor(
sample_interval_seconds=0.001,
recent_sample_count=4,
)
monitor.start()
task = monitor._task
monitor.start()
assert monitor._task is task
await asyncio.sleep(0.005)
assert monitor.snapshot()['samples_total'] > 0
assert monitor.snapshot()['running'] is True
await monitor.stop()
await monitor.stop()
assert monitor.snapshot()['running'] is False
assert task is not None and task.done()
async def test_event_loop_monitor_observes_real_scheduler_stall() -> None:
monitor = EventLoopLagMonitor(
sample_interval_seconds=0.005,
recent_sample_count=8,
)
monitor.start()
try:
await asyncio.sleep(0.01)
time.sleep(0.05)
await asyncio.sleep(0.01)
assert monitor.snapshot()['recent_max_lag_ms'] >= 35
finally:
await monitor.stop()
+134
View File
@@ -6,8 +6,14 @@ Tests session management, reuse, and cleanup.
from __future__ import annotations
import asyncio
import threading
from types import SimpleNamespace
from unittest.mock import AsyncMock
import pytest
import aiohttp
import httpx
from aiohttp import web
from langbot.pkg.utils import httpclient
@@ -25,6 +31,7 @@ class TestGetSession:
assert isinstance(session, aiohttp.ClientSession)
assert not session.closed
assert isinstance(session.cookie_jar, aiohttp.DummyCookieJar)
# Cleanup
await session.close()
@@ -87,6 +94,124 @@ class TestCloseAll:
assert len(httpclient._sessions) == 0
class TestReadLimited:
async def test_rejects_oversized_content_length_before_reading(self):
content = SimpleNamespace(iter_chunked=None)
response = SimpleNamespace(headers={'Content-Length': '11'}, content=content)
with pytest.raises(httpclient.RemoteResponseTooLargeError):
await httpclient.read_limited(response, max_bytes=10)
async def test_rejects_chunked_body_that_crosses_limit(self):
class Content:
async def iter_chunked(self, _chunk_size):
yield b'12345'
yield b'678901'
response = SimpleNamespace(headers={}, content=Content())
with pytest.raises(httpclient.RemoteResponseTooLargeError):
await httpclient.read_limited(response, max_bytes=10)
async def test_returns_body_within_limit(self):
class Content:
async def iter_chunked(self, _chunk_size):
yield b'12345'
yield b'67890'
response = SimpleNamespace(headers={}, content=Content())
assert await httpclient.read_limited(response, max_bytes=10) == b'1234567890'
async def test_json_reader_uses_same_limit(self):
class Content:
async def iter_chunked(self, _chunk_size):
yield b'{"ok":true}'
response = SimpleNamespace(
headers={},
content=Content(),
)
assert await httpclient.read_json_limited(response, max_bytes=16) == {'ok': True}
async def test_response_json_parse_runs_off_event_loop(self):
event_loop_thread = threading.get_ident()
response = SimpleNamespace(json=lambda: threading.get_ident())
assert await httpclient.parse_json_response(response) != event_loop_thread
async def test_response_json_parse_supports_async_test_doubles(self):
response = SimpleNamespace(json=AsyncMock(return_value={'ok': True}))
assert await httpclient.parse_json_response(response) == {'ok': True}
async def test_response_text_runs_off_loop_and_caps_diagnostics(self):
event_loop_thread = threading.get_ident()
class Response:
@property
def text(self):
return f'{threading.get_ident()}:abcdef'
value = await httpclient.response_text(Response(), max_chars=4)
assert not value.startswith(str(event_loop_thread))
assert value.endswith('[truncated]')
async def test_httpx_hook_rejects_before_automatic_buffer_grows(self):
class Source(httpx.AsyncByteStream):
def __init__(self):
self.closed = False
async def __aiter__(self):
yield b'123'
yield b'45'
async def aclose(self):
self.closed = True
source = Source()
transport = httpx.MockTransport(lambda _request: httpx.Response(200, stream=source))
async with httpx.AsyncClient(
transport=transport,
event_hooks=httpclient.httpx_response_limit_hooks(max_bytes=4),
) as client:
with pytest.raises(httpclient.RemoteResponseTooLargeError, match='4-byte'):
await client.get('https://example.invalid')
assert source.closed
async def test_httpx_limited_stream_closes_source_when_consumer_is_cancelled(self):
class Source(httpx.AsyncByteStream):
def __init__(self):
self.closed = False
async def __aiter__(self):
yield b'123'
await asyncio.Event().wait()
async def aclose(self):
self.closed = True
source = Source()
stream = httpclient._LimitedHTTPXAsyncByteStream(source, max_bytes=4)
first_chunk_consumed = asyncio.Event()
async def consume():
async for _chunk in stream:
first_chunk_consumed.set()
task = asyncio.create_task(consume())
await first_chunk_consumed.wait()
task.cancel()
with pytest.raises(asyncio.CancelledError):
await task
assert source.closed
async def test_close_all_handles_already_closed(self):
"""close_all handles already closed sessions gracefully."""
session = httpclient.get_session()
@@ -144,3 +269,12 @@ class TestSessionPoolIntegration:
assert session is session2
await httpclient.close_all()
async def test_shared_session_does_not_persist_cookies(self):
"""Shared transport pooling never creates cross-Workspace cookie state."""
session = httpclient.get_session()
session.cookie_jar.update_cookies({'workspace_session': 'secret'})
assert list(session.cookie_jar) == []
await httpclient.close_all()
+17
View File
@@ -10,11 +10,28 @@ import pytest
import base64
from langbot.pkg.utils.image import (
decode_base64_limited,
encode_base64,
get_qq_image_downloadable_url,
extract_b64_and_format,
)
@pytest.mark.asyncio
async def test_base64_media_helpers_round_trip_within_limit():
encoded = await encode_base64(b'1234')
assert await decode_base64_limited(encoded, max_bytes=4) == b'1234'
@pytest.mark.asyncio
async def test_base64_media_decode_rejects_oversized_payload():
encoded = base64.b64encode(b'12345').decode()
with pytest.raises(ValueError, match='exceeds'):
await decode_base64_limited(encoded, max_bytes=4)
class TestGetQQImageDownloadableUrl:
"""Tests for get_qq_image_downloadable_url function."""
+15 -1
View File
@@ -7,7 +7,13 @@ Tests log page management and pointer-based retrieval.
from __future__ import annotations
from langbot.pkg.utils.logcache import LogPage, LogCache, LOG_PAGE_SIZE, MAX_CACHED_PAGES
from langbot.pkg.utils.logcache import (
LogPage,
LogCache,
LOG_PAGE_SIZE,
MAX_CACHED_PAGES,
MAX_LOG_LINE_CHARS,
)
class TestLogPage:
@@ -208,3 +214,11 @@ class TestLogCache:
"""LOG_PAGE_SIZE is defined and reasonable."""
assert LOG_PAGE_SIZE > 0
assert LOG_PAGE_SIZE <= 1000 # Reasonable upper bound
def test_single_log_line_is_bounded(self):
cache = LogCache()
cache.add_log('x' * (MAX_LOG_LINE_CHARS * 2))
assert len(cache.log_pages[0].logs[0]) == MAX_LOG_LINE_CHARS
assert cache.log_pages[0].logs[0].endswith('[log truncated]')
+73
View File
@@ -0,0 +1,73 @@
from __future__ import annotations
import threading
import pytest
from langbot.pkg.utils import safe_regex
from langbot.pkg.utils.bounded_executor import blocking_work_scope, current_blocking_work_scope
@pytest.mark.asyncio
async def test_matches_any_runs_off_event_loop_and_preserves_workspace_scope(monkeypatch):
event_loop_thread = threading.get_ident()
observed: dict[str, object] = {}
original = safe_regex._matches_any_sync
def observe(*args, **kwargs):
observed['thread'] = threading.get_ident()
observed['scope'] = current_blocking_work_scope()
return original(*args, **kwargs)
monkeypatch.setattr(safe_regex, '_matches_any_sync', observe)
with blocking_work_scope('workspace-a'):
assert await safe_regex.matches_any(['^hello'], 'hello world') is True
assert observed['scope'] == 'workspace-a'
assert observed['thread'] != event_loop_thread
@pytest.mark.asyncio
async def test_matches_any_interrupts_catastrophic_backtracking():
with pytest.raises(safe_regex.SafeRegexTimeoutError):
await safe_regex.matches_any(
[r'(a+)+$'],
('a' * 100_000) + '!',
timeout_seconds=0.001,
)
@pytest.mark.asyncio
async def test_matches_any_rejects_pattern_and_input_amplification():
with pytest.raises(safe_regex.SafeRegexLimitError):
await safe_regex.matches_any(
['a'] * (safe_regex.MAX_PATTERN_COUNT + 1),
'a',
)
with pytest.raises(safe_regex.SafeRegexLimitError):
await safe_regex.matches_any(
['a'],
'a' * (safe_regex.MAX_INPUT_CHARS + 1),
)
@pytest.mark.asyncio
async def test_mask_patterns_bounds_replacement_growth_and_masks_matches():
found, masked = await safe_regex.mask_patterns(
[r'secret-\d+'],
'a secret-42 value',
mask='*',
mask_word='[hidden]',
)
assert found is True
assert masked == 'a [hidden] value'
with pytest.raises(safe_regex.SafeRegexLimitError):
await safe_regex.mask_patterns(
['a'],
'a' * safe_regex.MAX_INPUT_CHARS,
mask='0123456789',
mask_word='',
)