Files
GarminHealthLab/backend/tests/test_scheduler.py
ericwyuan 41b7ae82e4 fix(sync): 「全部历史」真的拉全部历史,自动同步不再每次静默失败
四个独立的 bug 叠在一起,表现为「只同步两天、没有进度」:

* 前端 `...(days ? { days } : {})` 把 days=0 当成未传。「全部历史」
  存的就是 0,请求体里根本没有 days,后端退回 7 天默认值。
* scheduler 用 `s[0]` 读 query_one 返回的 dict,抛 KeyError 后被
  per-account 的 except 吞掉。只要用户存过一次设置,每 30 分钟的
  自动同步就一次都没成功过——库里那 2 天全是手动点出来的。
* 增量同步查 `health_daily`(表其实叫 health_data),后台线程直接
  死掉,状态永远卡在 syncing,进度条不动。
* UI 完全不看 /sync 的返回值,rate_limited 时按钮点了没反应;轮询
  结束时又把 rate_limited 归进 else 分支报「同步完成」。

顺带:
* 日循环遇到 429 立即退避并保留已拉到的天数,而不是当成「跳过一天」
  继续往下捶 700 天——这正是之前限流死循环的来源之一。
* 定时循环显式传 SYNC_DAYS。历史范围按 UI 文案只描述手动全量同步,
  让半小时一次的 tick 重拉 730 天必然把限流撞得更深。
* 短同步逐天上报进度(原来每 5 天一次,7 天的同步全程停在 0)。
* /sync 路由重复解析 body,空 body 会 None.get 崩。
* 4 个 StubGarth 缺 configure(),7 个测试在此之前一直是红的。

新增 deploy/push.sh:NAS 只认密码,脚本开一个 ssh 复用连接,密码只
输一次,后面推送 / 重启 / 健康检查全走它。不碰 .env、.venv 和数据库。

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-01 08:11:24 +08:00

242 lines
8.6 KiB
Python

"""
Unit tests for the auto-sync scheduler.
The interesting behaviour is the claim: gunicorn runs several workers, each of
which starts its own timer, so without coordination one hourly tick would fire
a sync per worker.
"""
import datetime
import pytest
from services import scheduler
@pytest.fixture(autouse=True)
def _clean(db):
db.execute("DELETE FROM job_locks")
def set_last_run(db, when):
db.execute(
"UPDATE job_locks SET last_run_at = ? WHERE name = ?",
[when.isoformat(timespec="seconds"), scheduler.JOB_NAME],
)
class TestClaim:
def test_first_caller_gets_the_job(self, db):
assert scheduler.claim() is True
def test_second_caller_is_turned_away_while_the_first_holds_it(self, db, monkeypatch):
monkeypatch.setattr(scheduler.os, "getpid", lambda: 111)
assert scheduler.claim() is True
# A different worker, same instant.
monkeypatch.setattr(scheduler.os, "getpid", lambda: 222)
assert scheduler.claim() is False, "two workers must not run the same tick"
def test_not_due_again_within_the_interval(self, db):
scheduler.claim()
scheduler.release()
assert scheduler.claim() is False
def test_due_again_after_the_interval(self, db):
scheduler.claim()
scheduler.release()
set_last_run(db, datetime.datetime.utcnow() - datetime.timedelta(seconds=7200))
assert scheduler.claim() is True
def test_abandoned_claim_expires(self, db, monkeypatch):
"""A worker that dies mid-run must not block the job forever."""
monkeypatch.setattr(scheduler.os, "getpid", lambda: 111)
scheduler.claim()
stale = datetime.datetime.utcnow() - datetime.timedelta(
seconds=scheduler.CLAIM_TIMEOUT_SECONDS + 60
)
db.execute(
"UPDATE job_locks SET claimed_at = ? WHERE name = ?",
[stale.isoformat(timespec="seconds"), scheduler.JOB_NAME],
)
monkeypatch.setattr(scheduler.os, "getpid", lambda: 222)
assert scheduler.claim() is True
def test_release_without_running_leaves_it_due(self, db):
scheduler.claim()
scheduler.release(ran=False)
assert scheduler.claim() is True
def test_release_after_running_records_the_time(self, db):
scheduler.claim()
scheduler.release()
assert scheduler.status()["lastRunAt"] is not None
class TestSyncAllAccounts:
def test_no_accounts_is_a_no_op(self, db, user):
assert scheduler.sync_all_accounts() == []
def test_syncs_every_account_holding_a_token(self, db, user, monkeypatch, make_user):
from services import garmin as garmin_svc
other = make_user("b@example.com")
garmin_svc.save_token(user["id"], "t1", "a@example.com")
garmin_svc.save_token(other["id"], "t2", "b@example.com")
seen = []
monkeypatch.setattr(
garmin_svc, "sync_data",
lambda uid, creds, days=None, client=None: seen.append(uid)
or {"status": "success", "recordsSynced": days},
)
results = scheduler.sync_all_accounts(days=2)
assert set(seen) == {user["id"], other["id"]}
assert all(r["status"] == "success" for r in results)
def test_one_failing_account_does_not_stop_the_others(
self, db, user, monkeypatch, make_user
):
from services import garmin as garmin_svc
other = make_user("c@example.com")
garmin_svc.save_token(user["id"], "t1")
garmin_svc.save_token(other["id"], "t2")
def flaky(uid, creds, days=None, client=None):
if uid == user["id"]:
raise RuntimeError("token expired")
return {"status": "success", "recordsSynced": 2}
monkeypatch.setattr(garmin_svc, "sync_data", flaky)
results = scheduler.sync_all_accounts()
assert len(results) == 2
assert {r["status"] for r in results} == {"error", "success"}
def test_accounts_without_a_token_are_skipped(self, db, user, monkeypatch):
from services import garmin as garmin_svc
called = []
monkeypatch.setattr(
garmin_svc, "sync_data",
lambda *a, **k: called.append(1) or {"status": "success"},
)
scheduler.sync_all_accounts()
assert called == [], "an account with no token cannot be synced"
class TestStatus:
def test_reports_configuration(self, db):
s = scheduler.status()
assert s["intervalSeconds"] == scheduler.INTERVAL_SECONDS
assert s["days"] == scheduler.SYNC_DAYS
def test_next_run_follows_the_last(self, db):
scheduler.claim()
scheduler.release()
s = scheduler.status()
assert s["lastRunAt"] and s["nextRunAt"]
assert s["nextRunAt"] > s["lastRunAt"]
def test_no_run_yet(self, db):
s = scheduler.status()
assert s["lastRunAt"] is None
assert s["nextRunAt"] is None
class TestEndpoints:
def test_sync_latest_requires_auth(self, client):
assert client.post("/api/garmin/sync-latest", json={}).status_code == 401
def test_sync_latest_needs_a_bound_account(self, client, auth):
r = client.post("/api/garmin/sync-latest", headers=auth, json={})
assert r.status_code == 400
assert "绑定" in r.get_json()["error"]
def test_sync_latest_runs_inline(self, client, auth, user, db, monkeypatch):
from services import garmin as garmin_svc
garmin_svc.save_token(user["id"], "blob")
monkeypatch.setattr(
garmin_svc, "sync_data",
lambda uid, creds, days=None, client=None: {
"status": "success", "recordsSynced": days, "message": "ok",
},
)
r = client.post("/api/garmin/sync-latest", headers=auth, json={"days": 3})
assert r.status_code == 200
assert r.get_json()["recordsSynced"] == 3
def test_sync_latest_window_is_clamped(self, client, auth, user, db, monkeypatch):
from services import garmin as garmin_svc
garmin_svc.save_token(user["id"], "blob")
monkeypatch.setattr(
garmin_svc, "sync_data",
lambda uid, creds, days=None, client=None: {
"status": "success", "recordsSynced": days,
},
)
r = client.post("/api/garmin/sync-latest", headers=auth, json={"days": 999})
assert r.get_json()["recordsSynced"] == 7
def test_auto_sync_status_endpoint(self, client, auth):
r = client.get("/api/garmin/auto-sync", headers=auth)
assert r.status_code == 200
assert "intervalSeconds" in r.get_json()
class TestSyncWindowResolution:
"""Which window each account is synced over.
`query_one` returns a dict, so reading the saved 历史范围 as `s[0]` raised
KeyError — swallowed by the per-account handler, which meant every
scheduled tick failed for every account that had ever opened 设置, and
nothing was synced automatically at all.
"""
def _account(self, user, monkeypatch):
from services import garmin as garmin_svc
garmin_svc.save_token(user["id"], "t1")
seen = {}
def record(uid, creds, days=None, client=None):
seen["days"] = days
return {"status": "success", "recordsSynced": 0}
monkeypatch.setattr(garmin_svc, "sync_data", record)
return seen
def test_saved_history_days_is_honoured(self, db, user, monkeypatch):
from services import settings as settings_svc
seen = self._account(user, monkeypatch)
settings_svc.save_settings(user["id"], {"historyDays": 90})
results = scheduler.sync_all_accounts()
assert [r["status"] for r in results] == ["success"]
assert seen["days"] == 90
def test_full_history_becomes_the_maximum(self, db, user, monkeypatch):
from services import settings as settings_svc
seen = self._account(user, monkeypatch)
settings_svc.save_settings(user["id"], {"historyDays": 0})
scheduler.sync_all_accounts()
assert seen["days"] == 730
def test_the_scheduled_tick_ignores_the_history_setting(
self, db, user, monkeypatch
):
"""历史范围 is the window for the manual 全量同步. A half-hourly tick
re-pulling 全部历史 is what kept the account in a 429 loop."""
from services import settings as settings_svc
seen = self._account(user, monkeypatch)
settings_svc.save_settings(user["id"], {"historyDays": 0, "autoSync": True})
scheduler.sync_all_accounts(days=scheduler.SYNC_DAYS, respect_schedule=True)
assert seen["days"] == scheduler.SYNC_DAYS