feat(worker): ручной пересчёт метрик через очередь, диагностика шагов и heartbeat

POST /metrics/refresh ставит задачу в sync_job (source=METRICS_JOB) и отвечает 202, пересчёт делает worker; GET /metrics/status отдаёт refreshing и consistent. metric_refresh_log хранит failed_step и step_timings. Источники с needs="tinvest_token" не попадают в расписание без токена, tinvest/moex добавлены в default_schedule. Воркер трогает heartbeat-файл для healthcheck.
This commit is contained in:
Dmitry
2026-09-19 21:54:30 +03:00
parent 989a780f9f
commit 3b3ee4d682
11 changed files with 341 additions and 30 deletions
+95 -8
View File
@@ -1,8 +1,19 @@
from datetime import timedelta
from sqlalchemy import select
from factories import make_account, make_category, make_txn, month_back
from fintracker.analytics import today_local
from fintracker.models import AccountRole
from fintracker.models import METRICS_JOB, AccountRole, JobStatus, SyncJob
from fintracker.worker.scheduler import _process_queued_jobs
async def _queue_and_run(client, auth_headers):
"""POST /metrics/refresh, then let the worker take the job, as it does every 5 s."""
r = await client.post("/api/v1/metrics/refresh", headers=auth_headers)
assert r.status_code == 202, r.text
await _process_queued_jobs()
return r.json()
async def test_refresh_populates_every_metric_endpoint(client, auth_headers):
@@ -20,15 +31,20 @@ async def test_refresh_populates_every_metric_endpoint(client, auth_headers):
)
await make_txn(today_local() - timedelta(days=1), outcome="500", outcome_account_id=card)
assert (await client.get("/api/v1/metrics/status", headers=auth_headers)).json() is None
before = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
assert before == {"last_refresh": None, "consistent": True, "refreshing": False}
r = await client.post("/api/v1/metrics/refresh", headers=auth_headers)
assert r.status_code == 202
assert r.json()["error"] is None
assert r.json()["finished_at"] is not None
job = await _queue_and_run(client, auth_headers)
assert job["status"] == "queued"
status = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
assert status["trigger"] == "manual"
assert status["last_refresh"]["trigger"] == "manual"
assert status["last_refresh"]["error"] is None
assert status["last_refresh"]["finished_at"] is not None
assert status["last_refresh"]["failed_step"] is None
assert status["last_refresh"]["step_timings"] # every step is timed
assert status["consistent"] is True
assert status["refreshing"] is False
series = (await client.get("/api/v1/networth/series", headers=auth_headers)).json()
assert series
@@ -81,7 +97,7 @@ async def test_spending_rejects_a_bad_month(client, auth_headers):
async def test_empty_database_reports_no_transactions(client, auth_headers):
await client.post("/api/v1/metrics/refresh", headers=auth_headers)
await _queue_and_run(client, auth_headers)
quality = (await client.get("/api/v1/data-quality", headers=auth_headers)).json()
assert [row["check_name"] for row in quality] == ["no_transactions"]
assert (await client.get("/api/v1/runway", headers=auth_headers)).json()[
@@ -89,3 +105,74 @@ async def test_empty_database_reports_no_transactions(client, auth_headers):
] is None
assert (await client.get("/api/v1/networth/series", headers=auth_headers)).json() == []
assert (await client.get("/api/v1/spending/categories", headers=auth_headers)).json() == []
async def test_refresh_returns_before_the_work_is_done(client, auth_headers):
"""The 202 is honest: nothing has been rebuilt until the worker picks the job up."""
r = await client.post("/api/v1/metrics/refresh", headers=auth_headers)
assert r.status_code == 202
assert r.json()["source"] == METRICS_JOB
status = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
assert status["last_refresh"] is None
assert status["refreshing"] is True
await _process_queued_jobs()
status = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
assert status["refreshing"] is False
assert status["last_refresh"] is not None
async def test_a_waiting_refresh_is_shared_not_duplicated(client, auth_headers):
first = (await client.post("/api/v1/metrics/refresh", headers=auth_headers)).json()
second = (await client.post("/api/v1/metrics/refresh", headers=auth_headers)).json()
assert first["id"] == second["id"]
from fintracker.db import get_sessionmaker
async with get_sessionmaker()() as session:
jobs = (
(await session.execute(select(SyncJob).where(SyncJob.source == METRICS_JOB)))
.scalars()
.all()
)
assert len(jobs) == 1
async def test_a_refresh_that_fails_part_way_is_reported_as_inconsistent(
client, auth_headers, monkeypatch
):
from fintracker.db import get_sessionmaker
from fintracker.metrics import refresh as refresh_module
async def fine(session):
pass
async def boom(session):
raise RuntimeError("step blew up")
monkeypatch.setattr("fintracker.analytics.register_steps", lambda: None)
monkeypatch.setattr(refresh_module, "STEPS", [("fine", fine), ("boom", boom), ("never", fine)])
await _queue_and_run(client, auth_headers)
status = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
last = status["last_refresh"]
assert last["failed_step"] == "boom"
assert "step blew up" in last["error"]
assert list(last["step_timings"]) == ["fine"] # 'never' did not run, 'boom' did not finish
assert status["consistent"] is False
async with get_sessionmaker()() as session:
job = (
await session.execute(select(SyncJob).where(SyncJob.source == METRICS_JOB))
).scalar_one()
assert job.status == JobStatus.error
# a later clean run makes the tables one snapshot again
monkeypatch.setattr(refresh_module, "STEPS", [("fine", fine)])
await _queue_and_run(client, auth_headers)
status = (await client.get("/api/v1/metrics/status", headers=auth_headers)).json()
assert status["consistent"] is True
assert status["last_refresh"]["failed_step"] is None
+49
View File
@@ -0,0 +1,49 @@
"""The timetable only names sources that exist, and leaves off the ones that cannot run."""
from __future__ import annotations
from fintracker.config import Settings
from fintracker.sources import registry
from fintracker.worker.jobs import default_schedule, scheduled
def test_every_scheduled_source_is_registered_once():
names = [s.source for s in default_schedule()]
assert len(names) == len(set(names))
assert set(names) <= set(registry.names())
def test_every_live_source_is_on_the_timetable():
assert {s.source for s in default_schedule()} == set(registry.names())
def test_sources_needing_a_token_are_left_off_without_one():
ready, skipped = scheduled(Settings.model_construct(tinvest_token=None))
assert {s.source for s in skipped} == {"tinvest", "tinvest_events"}
assert not {s.source for s in ready} & {"tinvest", "tinvest_events"}
ready, skipped = scheduled(Settings.model_construct(tinvest_token="t.x"))
assert skipped == []
assert {"tinvest", "moex", "tinvest_events", "moex_payouts"} <= {s.source for s in ready}
def test_needs_names_a_real_setting():
for spec in default_schedule():
if spec.needs is not None:
assert spec.needs in Settings.model_fields
def test_worker_timetable_has_the_heartbeat_and_the_queue_poller():
from fintracker.worker.scheduler import build_scheduler
ids = {j.id for j in build_scheduler().get_jobs()}
assert {"heartbeat", "poll-sync-jobs", "sync:zenmoney", "sync:moex", "sync:moex_payouts"} <= ids
async def test_heartbeat_touches_its_file(tmp_path, monkeypatch):
from fintracker.worker import scheduler
beat = tmp_path / "beat"
monkeypatch.setattr(scheduler, "HEARTBEAT_FILE", beat)
await scheduler._heartbeat()
assert beat.exists()