mirror of
https://github.com/fawney19/Aether.git
synced 2026-09-03 01:40:21 +08:00
perf: 依赖数据库 CASCADE/SET NULL 替代手动清理关联表，缩短删除事务
- 批量删除移除 cleanup_key_references 手动清理，改为依赖 FK CASCADE/SET NULL - video_tasks.key_id FK 增加 ondelete="SET NULL"，附带幂等迁移脚本 - _sync_delete 增加 statement_timeout 和任务级超时保护 - 批量导入在每次 await 前释放闲置 DB 连接，按批次提交写入避免长事务 - 前端轮询改为先查后等，首次查询不再多等一个间隔
This commit is contained in:
fawney19
@@ -8,12 +8,14 @@ from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import json
|
||||
import time
|
||||
import uuid
|
||||
from collections.abc import Callable
|
||||
from concurrent.futures import Future
|
||||
|
||||
import redis.asyncio as aioredis
|
||||
from sqlalchemy import delete as sa_delete
|
||||
from sqlalchemy import text
|
||||
|
||||
from src.clients.redis_client import get_redis_client
|
||||
from src.core.logger import logger
|
||||
@@ -30,6 +32,12 @@ _TASK_RETAIN_SECONDS = 600
|
||||
# 批量删除时每个独立事务处理的 key 数量
|
||||
_CLEANUP_BATCH_SIZE = 50
|
||||
|
||||
# 单个批次的数据库 statement 超时（秒）
|
||||
_BATCH_STATEMENT_TIMEOUT_S = 30
|
||||
|
||||
# 整个任务的最大执行时间（秒）
|
||||
_TASK_TIMEOUT_S = 600
|
||||
|
||||
# Redis key 前缀
|
||||
_REDIS_KEY_PREFIX = "batch_delete_task"
|
||||
|
||||
@@ -175,23 +183,35 @@ def _sync_delete(
|
||||
) -> int:
|
||||
"""在线程中执行的同步删除逻辑，避免阻塞事件循环。
|
||||
|
||||
按小批次（_CLEANUP_BATCH_SIZE）清理关联表并删除 key，
|
||||
每个批次独立事务，单批失败跳过并继续，确保进度条持续推进。
|
||||
按小批次（_CLEANUP_BATCH_SIZE）直接删除 key，依赖数据库 CASCADE/SET NULL 自动清理关联表。
|
||||
每个批次独立事务，单批失败跳过并继续。
|
||||
"""
|
||||
from src.database import create_session
|
||||
from src.models.database import ProviderAPIKey
|
||||
from src.services.provider_keys.key_side_effects import cleanup_key_references
|
||||
|
||||
db = create_session()
|
||||
try:
|
||||
affected = 0
|
||||
total_batches = (len(key_ids) + _CLEANUP_BATCH_SIZE - 1) // _CLEANUP_BATCH_SIZE
|
||||
batch_idx = 0
|
||||
task_start = time.monotonic()
|
||||
for i in range(0, len(key_ids), _CLEANUP_BATCH_SIZE):
|
||||
# 整体超时保护
|
||||
if time.monotonic() - task_start > _TASK_TIMEOUT_S:
|
||||
logger.warning(
|
||||
"[BATCH_DELETE] task timeout after {}s, deleted {}/{}",
|
||||
_TASK_TIMEOUT_S,
|
||||
affected,
|
||||
len(key_ids),
|
||||
)
|
||||
break
|
||||
|
||||
batch = key_ids[i : i + _CLEANUP_BATCH_SIZE]
|
||||
batch_idx += 1
|
||||
try:
|
||||
cleanup_key_references(db, batch)
|
||||
# 设置 statement_timeout，防止单条 SQL 无限等锁
|
||||
timeout_ms = _BATCH_STATEMENT_TIMEOUT_S * 1000
|
||||
db.execute(text(f"SET LOCAL statement_timeout = '{timeout_ms}'"))
|
||||
result = db.execute(
|
||||
sa_delete(ProviderAPIKey).where(
|
||||
ProviderAPIKey.provider_id == provider_id,
|
||||
@@ -203,7 +223,9 @@ def _sync_delete(
|
||||
db.commit()
|
||||
except Exception as exc:
|
||||
logger.warning(
|
||||
"[BATCH_DELETE] batch failed (keys {}-{}): {}",
|
||||
"[BATCH_DELETE] batch {}/{} failed (keys {}-{}): {}",
|
||||
batch_idx,
|
||||
total_batches,
|
||||
i,
|
||||
i + len(batch),
|
||||
exc,
|
||||
@@ -212,9 +234,16 @@ def _sync_delete(
|
||||
db.rollback()
|
||||
except Exception:
|
||||
pass
|
||||
# 每 5 个批次或最后一个批次上报进度，避免过于频繁写 Redis
|
||||
if progress_callback is not None and (batch_idx % 5 == 0 or batch_idx == total_batches):
|
||||
# 每个批次都上报一次进度
|
||||
if progress_callback is not None:
|
||||
progress_callback(affected)
|
||||
logger.info(
|
||||
"[BATCH_DELETE] sync delete finished: deleted={}/{} batches={} elapsed={:.1f}s",
|
||||
affected,
|
||||
len(key_ids),
|
||||
total_batches,
|
||||
time.monotonic() - task_start,
|
||||
)
|
||||
return affected
|
||||
finally:
|
||||
try:
|
||||
@@ -230,6 +259,12 @@ async def _run_batch_delete(
|
||||
) -> None:
|
||||
r = await get_redis_client(require_redis=False)
|
||||
await _update_task_field(task_id, r=r, status=STATUS_RUNNING)
|
||||
logger.info(
|
||||
"[BATCH_DELETE_TASK] started task={} provider={} total={}",
|
||||
task_id,
|
||||
provider_id[:8],
|
||||
len(key_ids),
|
||||
)
|
||||
|
||||
# 从工作线程安全地触发 Redis 进度更新
|
||||
loop = asyncio.get_running_loop()
|
||||
@@ -245,7 +280,10 @@ async def _run_batch_delete(
|
||||
pass
|
||||
|
||||
try:
|
||||
affected = await asyncio.to_thread(_sync_delete, provider_id, key_ids, on_progress)
|
||||
affected = await asyncio.wait_for(
|
||||
asyncio.to_thread(_sync_delete, provider_id, key_ids, on_progress),
|
||||
timeout=_TASK_TIMEOUT_S + 30, # 留一点余量给 _sync_delete 内部超时
|
||||
)
|
||||
|
||||
if progress_futures:
|
||||
results = await asyncio.gather(
|
||||
@@ -292,6 +330,15 @@ async def _run_batch_delete(
|
||||
affected,
|
||||
)
|
||||
|
||||
except asyncio.TimeoutError:
|
||||
msg = f"task timeout after {_TASK_TIMEOUT_S + 30}s"
|
||||
await _update_task_field(task_id, r=r, status=STATUS_FAILED, message=msg)
|
||||
logger.error(
|
||||
"[BATCH_DELETE_TASK] {} task={} provider={}",
|
||||
msg,
|
||||
task_id,
|
||||
provider_id[:8],
|
||||
)
|
||||
except asyncio.CancelledError:
|
||||
await _update_task_field(
|
||||
task_id,
|
||||
|
||||
@@ -31,7 +31,6 @@ from src.services.provider.fingerprint import generate_fingerprint, normalize_fi
|
||||
from src.services.provider_keys.auth_type import normalize_auth_type
|
||||
from src.services.provider_keys.duplicate_check import check_duplicate_key
|
||||
from src.services.provider_keys.key_side_effects import (
|
||||
cleanup_key_references,
|
||||
run_create_key_side_effects,
|
||||
run_delete_key_side_effects,
|
||||
run_update_key_side_effects,
|
||||
@@ -519,12 +518,10 @@ async def batch_delete_endpoint_keys_response(db: Session, key_ids: list[str]) -
|
||||
# 收集受影响的 provider_id
|
||||
affected_provider_ids = {key.provider_id for key in keys if key.provider_id}
|
||||
|
||||
# 批量 SQL DELETE，一次提交
|
||||
# 批量 SQL DELETE，依赖数据库 CASCADE/SET NULL 自动清理关联表
|
||||
success_count = 0
|
||||
try:
|
||||
found_id_list = list(found_ids)
|
||||
# 先清理关联表，避免 CASCADE 级联删除超时
|
||||
cleanup_key_references(db, found_id_list)
|
||||
db.execute(sa_delete(ProviderAPIKey).where(ProviderAPIKey.id.in_(found_id_list)))
|
||||
db.commit()
|
||||
success_count = len(found_ids)
|
||||
|
||||
Reference in New Issue
Block a user