fix: transplant PG init backfill rollback scope + audit-loop state upsert registration (code-to10 v2, from 854f810)

- db: translate INSERT OR IGNORE ... SELECT to ON CONFLICT DO NOTHING;
  register task_audit_loop_state (task_id) in _OR_REPLACE_PK (direct port).
- server: _backfill_default_quotas uses SAVEPOINT containment, no longer
  commits the caller's transaction (direct port; callers are init_db only).
- worker: NOT ported — integration _coord_child_rows has no column-fallback
  retry loop, so the InFailedSqlTransaction cascade site does not exist here.
- tests: rewritten for the integration base (no cloud.audit_loop /
  _coord_audit_loop_merge here); real-PG cases skip without DSN/driver.

(cherry picked from commit 79e6843610fcd275af874e7b4afc57af7ffa579e)
This commit is contained in:
host committed 2026-11-03 02:49:26 +00:00
1 parent 741eb387ed
commit 2dc35bd331
3 files changed
+341 -4

No files matched your search

+281
View File
@@ -0,0 +1,281 @@
"""PG init/backfill 回滚与审计状态 upsert 登记修复测试(集成基底重写版)。
移植来源:feat-pg-init-fix 分支提交 854f810 的
cloud/tests/test_pg_init_fix.py。原测试不可直接带,原因:
- 原测试 import main 上不存在的 cloud.audit_loop(LoopState、
save_loop_state),并调 worker._coord_audit_loop_merge;集成基底
(02835f6)没有这两个接口,照搬会 import 失败。
- 重写版只用集成基底上存在的接口:cloud.db._translate_sql、
cloud.db.connect、cloud.server._backfill_default_quotas、
cloud.server.init_db,以及 cloud/tests/test_postgres_backend.py
的 _has_real_driver / _real_pg_params(真实 PG 守卫口径与原
测试同款:须显式设置 CLOUD_TEST_PG_DSN 且装有 psycopg 驱动,
否则 skip,绝不 fail;DSN 只从环境变量读,测试不打印它)。
覆盖的移植内容(与原测试一一对应,落点不同处已注明):
- 问题一 (a):INSERT OR IGNORE ... SELECT 形在 PG 适配层下未
翻译,直接报 syntax error at or near "OR"。重写版直接断言
_translate_sql 的翻译结果(VALUES 形、SELECT 形含列 / 不含列
三种),不依赖真实 PG 即可回归。
- 问题一 (b):_backfill_default_quotas 的 except 整事务 rollback
把同事务里 migrate 的 ALTER 一并丢掉。SQLite 版断言保存点
containment(失败只回滚回填本身、成功不提前提交调用方事务);
真实 PG 版断言同款保存点语义 + 新鲜 init 后 tasks 仍有
paused_by_user / interrupted_reason(即 migrate 的 ALTER 未被
回填失败丢掉;原测试用 Worker 合流证明同一链条,基底无该
接口,改以 init 后列存在性直接证明)。
- 问题二:task_audit_loop_state 未登记进 _OR_REPLACE_PK,PG 下
INSERT OR REPLACE 退化为 ON CONFLICT DO NOTHING,已存在行的
round / last_decision 更新静默丢弃。重写版断言 _translate_sql
对该表产出 ON CONFLICT (task_id) DO UPDATE SET(含 round /
last_decision),即登记本身;SQLite 原生 OR REPLACE 的行更新
语义只作对照(证明“应有语义”长什么样,不证明 PG 行为);
PG 侧的等价行为由真实 PG 测试覆盖(无 DSN/驱动时 skip)。
Worker 部分说明:源提交的 worker 改动是给 _coord_child_rows 的
列回落重试(7 列 -> 6 列 -> 4 列)之间加 rollback,防 PG
InFailedSqlTransaction 连环;集成基底的 _coord_child_rows 只有
单次 4 列 SELECT、无回落重试、无 except 重查,级联发生的落点
不存在,故本移植不带 worker 改动(见摘要)。本文件不测 worker。
"""
import os
import sqlite3
import sys
import tempfile
import unittest
from pathlib import Path
from unittest import mock
try:
from cloud.db import _translate_sql, connect
from cloud.server import _backfill_default_quotas, init_db
except ImportError:
sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
from cloud.db import _translate_sql, connect
from cloud.server import _backfill_default_quotas, init_db
try:
from cloud.tests.test_postgres_backend import (
_has_real_driver,
_real_pg_params,
)
_PG_PARAMS = _real_pg_params()
except Exception:
_PG_PARAMS = None
def _pg_env():
"""返回 PG 测试用的环境变量 dict;无真实 PG 时返回 None。"""
if _PG_PARAMS is None or not _has_real_driver():
return None
host, port, dbname, user, password = _PG_PARAMS
return {
"CLOUD_DB_BACKEND": "postgres",
"CLOUD_DB_HOST": host,
"CLOUD_DB_PORT": str(port),
"CLOUD_DB_NAME": dbname,
"CLOUD_DB_USER": user,
"CLOUD_DB_PASSWORD": password,
}
class TranslateTestCase(unittest.TestCase):
def test_ignore_select_form_translated(self):
sql, kind = _translate_sql(
"INSERT OR IGNORE INTO user_quotas(user_id,"
" quota_usd_micros, updated_at)"
" SELECT u.id, ?, ? FROM users u"
" LEFT JOIN user_quotas q ON q.user_id = u.id"
" WHERE q.user_id IS NULL")
self.assertEqual(kind, "normal")
self.assertNotIn("OR IGNORE", sql)
self.assertIn("ON CONFLICT DO NOTHING", sql)
self.assertIn("%s", sql)
self.assertNotIn("?", sql)
def test_ignore_select_form_without_column_list_translated(self):
# 正则里列清单是可选组;无列清单的 SELECT 形同样要翻译,
# 不能原样透传到 PG 报 syntax error。
sql, kind = _translate_sql(
"INSERT OR IGNORE INTO user_quotas"
" SELECT u.id, ?, ? FROM users u"
" LEFT JOIN user_quotas q ON q.user_id = u.id"
" WHERE q.user_id IS NULL")
self.assertEqual(kind, "normal")
self.assertNotIn("OR IGNORE", sql)
self.assertIn("ON CONFLICT DO NOTHING", sql)
self.assertIn("SELECT u.id", sql)
def test_ignore_values_form_unchanged(self):
# 回归:既有 VALUES 形翻译不受新增 SELECT 形分支影响。
sql, kind = _translate_sql(
"INSERT OR IGNORE INTO user_quotas(user_id,"
" quota_usd_micros, updated_at) VALUES(?, ?, ?)")
self.assertEqual(kind, "normal")
self.assertNotIn("OR IGNORE", sql)
self.assertIn("ON CONFLICT DO NOTHING", sql)
def test_audit_loop_state_or_replace_updates(self):
sql, kind = _translate_sql(
"INSERT OR REPLACE INTO task_audit_loop_state"
"(task_id, round, passed_gates, criteria, last_decision,"
" updated_at) VALUES(?, ?, ?, ?, ?, ?)")
self.assertEqual(kind, "normal")
self.assertIn("ON CONFLICT (task_id) DO UPDATE SET", sql)
self.assertNotIn("DO NOTHING", sql)
self.assertIn("round = EXCLUDED.round", sql)
self.assertIn("last_decision = EXCLUDED.last_decision", sql)
class SqliteBackfillTestCase(unittest.TestCase):
def setUp(self):
self.tmp = tempfile.TemporaryDirectory()
self.db_path = os.path.join(self.tmp.name, "control.db")
def tearDown(self):
self.tmp.cleanup()
def _conn_with_users(self):
conn = sqlite3.connect(self.db_path)
conn.execute("CREATE TABLE users(id TEXT PRIMARY KEY)")
conn.execute("INSERT INTO users(id) VALUES('u1')")
conn.commit()
return conn
def test_backfill_does_not_commit_caller_transaction(self):
# 旧版成功时 conn.commit(),会把调用方未提交的改动一并
# 提交;改后提交由调用方负责,rollback 后调用方的改动
# 与回填行应一并消失。
conn = self._conn_with_users()
try:
conn.execute("INSERT INTO users(id) VALUES('u2')")
self.assertEqual(_backfill_default_quotas(conn), 2)
conn.rollback()
users = [r[0] for r in conn.execute(
"SELECT id FROM users ORDER BY id").fetchall()]
self.assertEqual(users, ["u1"])
# 回滚连 _ensure 建的 user_quotas 表一起丢掉(同事务),
# 表不在即等于回填行已消失;表若在则行数必为 0。
tables = [r[0] for r in conn.execute(
"SELECT name FROM sqlite_master"
" WHERE name = 'user_quotas'").fetchall()]
if tables:
quotas = conn.execute(
"SELECT COUNT(*) FROM user_quotas").fetchone()[0]
self.assertEqual(quotas, 0)
finally:
conn.close()
def test_backfill_failure_rolls_back_only_to_savepoint(self):
# user_quotas 缺 quota_usd_micros 列(_ensure 为
# CREATE TABLE IF NOT EXISTS,不会补列),INSERT 必失败;
# 旧版整事务 rollback 会把同事务里调用方的改动也丢掉,
# 保存点版只回滚回填本身。
conn = sqlite3.connect(self.db_path)
try:
conn.execute("CREATE TABLE users(id TEXT PRIMARY KEY)")
conn.execute(
"CREATE TABLE user_quotas(user_id TEXT PRIMARY KEY,"
" updated_at DOUBLE PRECISION)")
conn.execute("INSERT INTO users(id) VALUES('u1')")
conn.commit()
conn.execute("INSERT INTO users(id) VALUES('u2')")
self.assertEqual(_backfill_default_quotas(conn), 0)
users = [r[0] for r in conn.execute(
"SELECT id FROM users ORDER BY id").fetchall()]
self.assertEqual(users, ["u1", "u2"])
finally:
conn.close()
class SqliteUpsertSemanticsTestCase(unittest.TestCase):
def test_sqlite_native_replace_updates_existing_row(self):
# 对照用:SQLite 原生 OR REPLACE 的应有语义——同主键再写
# 一行,round / last_decision 被新值覆盖。PG 侧的等价行
# 为由 _OR_REPLACE_PK 登记保证,见 TranslateTestCase 与
# 真实 PG 测试;本用例不证明 PG 行为。
conn = sqlite3.connect(":memory:")
try:
conn.execute(
"CREATE TABLE task_audit_loop_state(task_id TEXT"
" PRIMARY KEY, round INTEGER, passed_gates TEXT,"
" criteria TEXT, last_decision TEXT,"
" updated_at DOUBLE PRECISION)")
conn.execute(
"INSERT OR REPLACE INTO task_audit_loop_state"
" VALUES('P1', 1, '[\"child:a\"]', '{}', 'pushback', 1.0)")
conn.execute(
"INSERT OR REPLACE INTO task_audit_loop_state"
" VALUES('P1', 2, '[\"child:a\"]', '{}', 'escalate', 2.0)")
conn.commit()
row = conn.execute(
"SELECT round, last_decision FROM task_audit_loop_state"
" WHERE task_id = 'P1'").fetchone()
self.assertEqual(row, (2, "escalate"))
finally:
conn.close()
class PgInitFixTestCase(unittest.TestCase):
"""真实 PG 测试:无 DSN/驱动时全部 skip。"""
def setUp(self):
env = _pg_env()
if env is None:
self.skipTest("未设置可达的 CLOUD_TEST_PG_DSN 或未装驱动,"
"跳过真实 PG 测试")
self._patcher = mock.patch.dict(os.environ, env)
self._patcher.start()
self.addCleanup(self._patcher.stop)
# 一次性库由外部建好并指向 DSN;init_db 直接跑,不做
# 任何夹具补偿(这正是本修复要证明的点)。
init_db(":memory:")
def _pg(self):
return connect(":memory:")
def test_pg_fresh_init_keeps_migrate_columns(self):
# 新鲜 PG 库 init:migrate_tasks_table 的 ALTER 与回填同
# 事务;旧代码下回填 SELECT 形未翻译报错、整事务 rollback
# 会把 ALTER 一并丢掉,tasks 缺 paused_by_user /
# interrupted_reason。改后两列必须在。
conn = self._pg()
try:
cols = {r[1] for r in conn.execute(
"PRAGMA table_info(tasks)").fetchall()}
self.assertIn("paused_by_user", cols)
self.assertIn("interrupted_reason", cols)
finally:
conn.close()
def test_pg_backfill_failure_preserves_caller_transaction(self):
# 与 SQLite 版同款保存点断言,跑在 PG 上:缺列导致
# INSERT 失败时,同事务里调用方的改动必须还在。
conn = self._pg()
try:
conn.execute("DROP TABLE IF EXISTS user_quotas")
conn.execute(
"CREATE TABLE user_quotas(user_id TEXT PRIMARY KEY,"
" updated_at DOUBLE PRECISION)")
conn.execute("CREATE TABLE IF NOT EXISTS users"
"(id TEXT PRIMARY KEY)")
conn.execute("INSERT INTO users(id) VALUES('u9')"
" ON CONFLICT DO NOTHING")
conn.commit()
conn.execute("INSERT INTO users(id) VALUES('u10')"
" ON CONFLICT DO NOTHING")
self.assertEqual(_backfill_default_quotas(conn), 0)
users = [r[0] for r in conn.execute(
"SELECT id FROM users WHERE id IN ('u9', 'u10')"
" ORDER BY id").fetchall()]
self.assertEqual(users, sorted(["u9", "u10"]))
conn.rollback()
finally:
conn.close()
if __name__ == "__main__":
unittest.main()