mirror of
https://github.com/tiennm99/DocsGPT.git
synced 2026-10-12 20:15:26 +00:00
Alex
574f96341e
The backend import package is now docsgpt, the name it will carry on PyPI; application was far too generic to install into anyone's site-packages. git mv plus a mechanical rewrite of every import, dotted string and path reference: 734 Python files, the compose files, Dockerfile, workflows, docs, setup scripts, devcontainer, k8s manifests, vscode config, pytest and coverage config, .gitignore. Behaviour is unchanged. Kept for one release: - A top-level application package whose meta-path finder resolves application.x.y to the already-imported docsgpt.x.y object, so old imports and entry points (celery -A application.app.celery, uvicorn application.asgi:asgi_app) keep working with a FutureWarning. - Celery registers every application.* task name as an alias of its docsgpt.* task on start-up, so messages queued by the previous release still run. The redbeat key prefix moves to redbeat:docsgpt:v2: so schedule entries the previous release wrote are left unread instead of firing twice. The backend image builds from the repository root (docker build -f docsgpt/Dockerfile .) so it can ship the alias package; a root .dockerignore allow-lists docsgpt/ and application/ and keeps caches, local data, .env files, the sample index files and the Dockerfile out. Compose and the image workflows point at the new context.
283 lines
10 KiB
Python
283 lines
10 KiB
Python
"""Tests for the scheduler reconciliation sweep + cleanup task."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from datetime import datetime, timedelta, timezone
|
|
|
|
import pytest
|
|
from sqlalchemy import text
|
|
|
|
from docsgpt.api.user.reconciliation import run_reconciliation
|
|
from docsgpt.storage.db.repositories.schedule_runs import (
|
|
ScheduleRunsRepository,
|
|
)
|
|
from docsgpt.storage.db.repositories.schedules import SchedulesRepository
|
|
|
|
|
|
def _now() -> datetime:
|
|
return datetime.now(timezone.utc)
|
|
|
|
|
|
def _make_pending_run(conn, *, user_id="u1"):
|
|
agent_id = str(
|
|
conn.execute(
|
|
text(
|
|
"INSERT INTO agents (user_id, name, status) "
|
|
"VALUES (:u, 'a', 'draft') RETURNING id"
|
|
),
|
|
{"u": user_id},
|
|
).fetchone()[0]
|
|
)
|
|
schedule = SchedulesRepository(conn).create(
|
|
user_id=user_id, agent_id=agent_id, trigger_type="recurring",
|
|
instruction="i", cron="*/5 * * * *",
|
|
next_run_at=_now() + timedelta(minutes=5),
|
|
)
|
|
run = ScheduleRunsRepository(conn).record_pending(
|
|
str(schedule["id"]), user_id, agent_id, _now(),
|
|
)
|
|
return schedule, run, agent_id
|
|
|
|
|
|
def _make_once_pending_run(conn, *, user_id="u1"):
|
|
"""Once-schedule + pending run variant of _make_pending_run."""
|
|
agent_id = str(
|
|
conn.execute(
|
|
text(
|
|
"INSERT INTO agents (user_id, name, status) "
|
|
"VALUES (:u, 'a', 'draft') RETURNING id"
|
|
),
|
|
{"u": user_id},
|
|
).fetchone()[0]
|
|
)
|
|
fire = _now() + timedelta(minutes=5)
|
|
schedule = SchedulesRepository(conn).create(
|
|
user_id=user_id, agent_id=agent_id, trigger_type="once",
|
|
instruction="do once", run_at=fire,
|
|
next_run_at=fire,
|
|
)
|
|
run = ScheduleRunsRepository(conn).record_pending(
|
|
str(schedule["id"]), user_id, agent_id, fire,
|
|
)
|
|
return schedule, run, agent_id
|
|
|
|
|
|
@pytest.fixture
|
|
def patched_engine(pg_engine, monkeypatch):
|
|
monkeypatch.setattr(
|
|
"docsgpt.api.user.reconciliation.get_engine",
|
|
lambda: pg_engine,
|
|
)
|
|
monkeypatch.setattr(
|
|
"docsgpt.api.user.reconciliation.settings",
|
|
type("S", (), {
|
|
"POSTGRES_URI": str(pg_engine.url),
|
|
"SCHEDULE_RUN_TIMEOUT": 60,
|
|
})(),
|
|
)
|
|
yield pg_engine
|
|
|
|
|
|
class TestReconciler:
|
|
def test_stuck_running_flipped_to_timeout(self, pg_engine, patched_engine):
|
|
with pg_engine.begin() as conn:
|
|
schedule, run, _ = _make_pending_run(conn)
|
|
ScheduleRunsRepository(conn).mark_running(run["id"], "t1")
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET started_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
summary = run_reconciliation()
|
|
assert summary["schedule_runs_failed"] >= 1
|
|
with pg_engine.connect() as conn:
|
|
row = ScheduleRunsRepository(conn).get_internal(run["id"])
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
assert row["status"] == "timeout"
|
|
assert row["error_type"] == "timeout"
|
|
assert sched["consecutive_failure_count"] == 1
|
|
|
|
def test_stuck_pending_flipped_to_failed(self, pg_engine, patched_engine):
|
|
"""A pending run whose worker never started reconciles to 'failed'."""
|
|
with pg_engine.begin() as conn:
|
|
schedule, run, _ = _make_pending_run(conn)
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET created_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
summary = run_reconciliation()
|
|
assert summary["schedule_runs_failed"] >= 1
|
|
with pg_engine.connect() as conn:
|
|
row = ScheduleRunsRepository(conn).get_internal(run["id"])
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
assert row["status"] == "failed"
|
|
assert row["error_type"] == "internal"
|
|
assert "worker_never_started" in (row["error"] or "")
|
|
assert sched["consecutive_failure_count"] == 1
|
|
|
|
def test_once_schedule_with_stuck_running_run_marked_completed(
|
|
self, pg_engine, patched_engine,
|
|
):
|
|
"""Once + stuck 'running' run -> parent flipped to 'completed'."""
|
|
with pg_engine.begin() as conn:
|
|
schedule, run, _ = _make_once_pending_run(conn)
|
|
ScheduleRunsRepository(conn).mark_running(run["id"], "t-once")
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET started_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
run_reconciliation()
|
|
with pg_engine.connect() as conn:
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
row = ScheduleRunsRepository(conn).get_internal(run["id"])
|
|
assert row["status"] == "timeout"
|
|
assert sched["status"] == "completed", (
|
|
"stuck once-run must terminal-flip the parent schedule"
|
|
)
|
|
assert sched["next_run_at"] is None
|
|
|
|
def test_once_schedule_with_stuck_pending_run_marked_completed(
|
|
self, pg_engine, patched_engine,
|
|
):
|
|
"""Once + stuck 'pending' run -> parent flipped to 'completed'."""
|
|
with pg_engine.begin() as conn:
|
|
schedule, run, _ = _make_once_pending_run(conn)
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET created_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
run_reconciliation()
|
|
with pg_engine.connect() as conn:
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
row = ScheduleRunsRepository(conn).get_internal(run["id"])
|
|
assert row["status"] == "failed"
|
|
assert sched["status"] == "completed", (
|
|
"stuck pending once-run must terminal-flip the parent schedule"
|
|
)
|
|
assert sched["next_run_at"] is None
|
|
|
|
def test_agentless_once_stuck_running_marked_completed(
|
|
self, pg_engine, patched_engine,
|
|
):
|
|
"""Stuck-run terminal flip works for agentless once-schedules."""
|
|
with pg_engine.begin() as conn:
|
|
fire = _now() + timedelta(minutes=5)
|
|
schedule = SchedulesRepository(conn).create(
|
|
user_id="u-agentless", agent_id=None, trigger_type="once",
|
|
instruction="agentless go", run_at=fire,
|
|
next_run_at=fire,
|
|
created_via="chat",
|
|
)
|
|
run = ScheduleRunsRepository(conn).record_pending(
|
|
str(schedule["id"]), "u-agentless", None, fire,
|
|
)
|
|
ScheduleRunsRepository(conn).mark_running(run["id"], "t-stuck")
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET started_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
run_reconciliation()
|
|
with pg_engine.connect() as conn:
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
row = ScheduleRunsRepository(conn).get_internal(run["id"])
|
|
assert row["status"] == "timeout"
|
|
assert sched["status"] == "completed"
|
|
assert sched["next_run_at"] is None
|
|
|
|
def test_recurring_schedule_with_stuck_run_stays_active(
|
|
self, pg_engine, patched_engine,
|
|
):
|
|
"""Recurring keeps firing; only the run flips, not the parent."""
|
|
with pg_engine.begin() as conn:
|
|
schedule, run, _ = _make_pending_run(conn)
|
|
ScheduleRunsRepository(conn).mark_running(run["id"], "t-rec")
|
|
conn.execute(
|
|
text(
|
|
"UPDATE schedule_runs "
|
|
"SET started_at = now() - interval '120 minutes' "
|
|
"WHERE id = CAST(:i AS uuid)"
|
|
),
|
|
{"i": run["id"]},
|
|
)
|
|
|
|
run_reconciliation()
|
|
with pg_engine.connect() as conn:
|
|
sched = SchedulesRepository(conn).get_internal(str(schedule["id"]))
|
|
assert sched["status"] == "active"
|
|
assert sched["consecutive_failure_count"] == 1
|
|
|
|
|
|
class TestCleanup:
|
|
def test_cleanup_schedule_runs_trims_old_rows(self, pg_engine, monkeypatch):
|
|
from docsgpt.api.user.tasks import cleanup_schedule_runs as _task
|
|
|
|
monkeypatch.setattr(
|
|
"docsgpt.storage.db.engine.get_engine",
|
|
lambda: pg_engine,
|
|
)
|
|
|
|
class S:
|
|
POSTGRES_URI = str(pg_engine.url)
|
|
SCHEDULE_RUN_OUTPUT_RETENTION_DAYS = 30
|
|
monkeypatch.setattr("docsgpt.api.user.tasks.settings", S, raising=False)
|
|
monkeypatch.setattr(
|
|
"docsgpt.core.settings.settings.POSTGRES_URI",
|
|
str(pg_engine.url),
|
|
raising=False,
|
|
)
|
|
monkeypatch.setattr(
|
|
"docsgpt.core.settings.settings.SCHEDULE_RUN_OUTPUT_RETENTION_DAYS",
|
|
30,
|
|
raising=False,
|
|
)
|
|
|
|
with pg_engine.begin() as conn:
|
|
schedule, _, _ = _make_pending_run(conn)
|
|
for i in range(60):
|
|
conn.execute(
|
|
text(
|
|
"""
|
|
INSERT INTO schedule_runs (
|
|
schedule_id, user_id, agent_id, status,
|
|
scheduled_for, created_at
|
|
) VALUES (
|
|
CAST(:s AS uuid), 'u1',
|
|
CAST(:a AS uuid), 'success',
|
|
now() - interval '100 days' - (:i * interval '1 second'),
|
|
now() - interval '100 days'
|
|
)
|
|
"""
|
|
),
|
|
{"s": str(schedule["id"]),
|
|
"a": str(schedule["agent_id"]),
|
|
"i": i},
|
|
)
|
|
|
|
result = _task.run()
|
|
assert isinstance(result.get("deleted"), int)
|
|
assert result["deleted"] >= 1
|