fix(worker): OSS上传崩溃修复 - connect超时 + 分片上传 + 总超时保护 #238

Merged
xiaoxia merged 1 commits from fix/oss-upload-crash into develop 2026-07-13 09:33:13 +08:00
2 changed files with 318 additions and 10 deletions
+82 -10
View File
@@ -9,6 +9,7 @@ from __future__ import annotations
import hashlib
import logging
import os
import threading
from pathlib import Path
from typing import Optional
from urllib.parse import urlparse
@@ -17,6 +18,13 @@ import oss2
logger = logging.getLogger(__name__)
# OSS 上传配置
OSS_CONNECT_TIMEOUT = 10 # 连接超时(秒),防止 TCP 握手挂死
OSS_UPLOAD_TOTAL_TIMEOUT = 300 # 单文件上传总超时(秒),防止网络慢时无限卡住
OSS_MULTIPART_THRESHOLD = 100 * 1024 * 1024 # 分片上传阈值:100MB 以上走分片
OSS_PART_SIZE = 8 * 1024 * 1024 # 分片大小:8MB
OSS_MULTIPART_NUM_THREADS = 3 # 分片上传并发数
# ── OSS 配置 ──────────────────────────────────────────────────────────────────
@@ -43,6 +51,9 @@ def oss_bucket() -> oss2.Bucket | None:
P0-2 修复:endpoint 不带 scheme 时自动补 https:// 前缀,
确保 sign_url 等依赖 scheme 的方法返回 HTTPS URL。
P0-staging 修复:增加 connect_timeout=10s,防止网络抖动时
TCP 握手阶段无限挂死,导致 worker 进程卡死。
Returns:
oss2.Bucket 实例,配置缺失时返回 None。
"""
@@ -53,7 +64,12 @@ def oss_bucket() -> oss2.Bucket | None:
# endpoint 无 scheme 时补 https://,与 API 端 storage.py 保持一致
if not endpoint.startswith(("http://", "https://")):
endpoint = f"https://{endpoint}"
return oss2.Bucket(oss2.Auth(access_key_id, access_key_secret), endpoint, bucket_name)
return oss2.Bucket(
oss2.Auth(access_key_id, access_key_secret),
endpoint,
bucket_name,
connect_timeout=OSS_CONNECT_TIMEOUT,
)
def normalize_storage_key(storage_key_or_url: str) -> str:
@@ -96,6 +112,9 @@ def download_asset(asset_storage_key: str, local_path: Path) -> bool:
def upload_to_oss(local_path: Path, storage_key: str) -> str | None:
"""上传文件到 OSS,返回公开 URL。
大文件(>100MB)自动走分片上传,降低内存峰值,减少 OOM 风险。
上传加总超时保护(默认 300s),防止网络异常时无限挂死。
Args:
local_path: 本地文件路径
storage_key: 目标存储键
@@ -106,18 +125,71 @@ def upload_to_oss(local_path: Path, storage_key: str) -> str | None:
bucket = oss_bucket()
if bucket is None:
return None
try:
bucket.put_object_from_file(storage_key, str(local_path))
settings = oss_settings()
if settings:
_, _, endpoint, bucket_name = settings
endpoint_clean = endpoint.replace("https://", "").replace("http://", "")
return f"https://{bucket_name}.{endpoint_clean}/{storage_key}"
result: dict = {"url": None, "error": None, "file_size": 0}
done = threading.Event()
def _do_upload():
try:
# 尝试获取文件大小,用于分片判断和日志;stat 失败时 fallback 走普通上传
try:
file_size = local_path.stat().st_size
result["file_size"] = file_size
use_multipart = file_size >= OSS_MULTIPART_THRESHOLD
except OSError:
use_multipart = False
file_size = 0
if use_multipart:
# 分片上传:降低内存峰值,每片 8MB,3 线程并发
logger.info(
"大文件分片上传: storage_key=%s, size=%.1fMB, part_size=%dMB, threads=%d",
storage_key[:80],
file_size / 1024 / 1024,
OSS_PART_SIZE // 1024 // 1024,
OSS_MULTIPART_NUM_THREADS,
)
oss2.resumable_upload(
bucket,
storage_key,
str(local_path),
multipart_threshold=OSS_MULTIPART_THRESHOLD,
part_size=OSS_PART_SIZE,
num_threads=OSS_MULTIPART_NUM_THREADS,
)
else:
bucket.put_object_from_file(storage_key, str(local_path))
# 构造返回 URL
settings = oss_settings()
if settings:
_, _, endpoint, bucket_name = settings
endpoint_clean = endpoint.replace("https://", "").replace("http://", "")
result["url"] = f"https://{bucket_name}.{endpoint_clean}/{storage_key}"
except Exception as e:
result["error"] = e
logger.exception("上传 OSS 失败: %s", storage_key)
finally:
done.set()
upload_thread = threading.Thread(target=_do_upload, daemon=True)
upload_thread.start()
finished = done.wait(timeout=OSS_UPLOAD_TOTAL_TIMEOUT)
if not finished:
logger.error(
"OSS 上传超时(%.0fs),强制中止: storage_key=%s, size=%.1fMB",
OSS_UPLOAD_TOTAL_TIMEOUT,
storage_key[:80],
result["file_size"] / 1024 / 1024 if result["file_size"] else 0,
)
return None
except Exception:
logger.exception("上传 OSS 失败: %s", storage_key)
if result["error"]:
return None
return result["url"]
def get_signed_download_url(storage_key_or_url: str, expires_seconds: int = 3600) -> str | None:
"""生成预签名下载 URL(用于私有 bucket 的 URL 校验或临时下载)。
+236
View File
@@ -0,0 +1,236 @@
"""P0-stagingOSS 上传崩溃修复测试.
测试:
1. oss_bucket() 传递 connect_timeout 参数
2. upload_to_oss() 小文件走 put_object_from_file,大文件走分片上传
3. upload_to_oss() 超时保护(超过总超时返回 None)
4. upload_to_oss() 异常时返回 None
"""
from __future__ import annotations
import os
import tempfile
import time
from pathlib import Path
from unittest.mock import MagicMock, patch
import pytest
# ── oss_bucket connect_timeout 测试 ───────────────────────────────────────────
class TestOSSBucketConnectTimeout:
"""测试 oss_bucket() 传递 connect_timeout 参数."""
def test_oss_bucket_has_connect_timeout(self):
"""oss_bucket 应传递 connect_timeout=10s 参数."""
from video_processing.oss_helpers import oss_bucket
mock_bucket_instance = MagicMock()
with (
patch.dict(
os.environ,
{
"OSS_ACCESS_KEY_ID": "test-key",
"OSS_ACCESS_KEY_SECRET": "test-secret",
"OSS_ENDPOINT": "oss-cn-hangzhou.aliyuncs.com",
"OSS_BUCKET_NAME": "test-bucket",
},
),
patch("video_processing.oss_helpers.oss2.Auth"),
patch("video_processing.oss_helpers.oss2.Bucket", return_value=mock_bucket_instance) as mock_bucket_cls,
):
bucket = oss_bucket()
assert bucket is mock_bucket_instance
# 验证 connect_timeout 关键字参数
call_kwargs = mock_bucket_cls.call_args[1]
assert "connect_timeout" in call_kwargs, "oss_bucket 应传递 connect_timeout 参数"
assert (
call_kwargs["connect_timeout"] == 10
), f"connect_timeout 应为 10,实际为 {call_kwargs['connect_timeout']}"
def test_oss_bucket_no_config_returns_none(self):
"""OSS 配置缺失时返回 None."""
from video_processing.oss_helpers import oss_bucket
with patch.dict(os.environ, {}, clear=True):
bucket = oss_bucket()
assert bucket is None
# ── upload_to_oss 分片上传测试 ────────────────────────────────────────────────
class TestUploadToOSSMultipart:
"""测试 upload_to_oss() 根据文件大小选择上传方式."""
def _create_temp_file(self, size_bytes: int) -> Path:
"""创建指定大小的临时文件."""
tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
tmp.write(b"x" * size_bytes)
tmp.close()
return Path(tmp.name)
def test_small_file_uses_put_object(self):
"""小文件(<100MB)走 put_object_from_file."""
from video_processing.oss_helpers import upload_to_oss
small_file = self._create_temp_file(10 * 1024 * 1024) # 10MB
try:
mock_bucket = MagicMock()
with (
patch.dict(
os.environ,
{
"OSS_ACCESS_KEY_ID": "test-key",
"OSS_ACCESS_KEY_SECRET": "test-secret",
"OSS_ENDPOINT": "oss-cn-hangzhou.aliyuncs.com",
"OSS_BUCKET_NAME": "test-bucket",
},
),
patch("video_processing.oss_helpers.oss2.Auth"),
patch("video_processing.oss_helpers.oss2.Bucket", return_value=mock_bucket),
patch("video_processing.oss_helpers.oss2.resumable_upload") as mock_resumable,
):
url = upload_to_oss(small_file, "test/small.mp4")
# 验证调用了 put_object_from_file
mock_bucket.put_object_from_file.assert_called_once()
# 验证没调用分片上传
mock_resumable.assert_not_called()
# 验证返回 URL
assert url == "https://test-bucket.oss-cn-hangzhou.aliyuncs.com/test/small.mp4"
finally:
small_file.unlink()
def test_large_file_uses_resumable_upload(self):
"""大文件(>=100MB)走 resumable_upload 分片上传."""
from video_processing.oss_helpers import upload_to_oss
large_file = self._create_temp_file(100 * 1024 * 1024) # 100MB
try:
mock_bucket = MagicMock()
with (
patch.dict(
os.environ,
{
"OSS_ACCESS_KEY_ID": "test-key",
"OSS_ACCESS_KEY_SECRET": "test-secret",
"OSS_ENDPOINT": "oss-cn-hangzhou.aliyuncs.com",
"OSS_BUCKET_NAME": "test-bucket",
},
),
patch("video_processing.oss_helpers.oss2.Auth"),
patch("video_processing.oss_helpers.oss2.Bucket", return_value=mock_bucket),
patch("video_processing.oss_helpers.oss2.resumable_upload") as mock_resumable,
):
url = upload_to_oss(large_file, "test/large.mp4")
# 验证调用了分片上传
mock_resumable.assert_called_once()
# 验证没调用 put_object_from_file
mock_bucket.put_object_from_file.assert_not_called()
# 验证分片参数
call_kwargs = mock_resumable.call_args[1]
assert call_kwargs["multipart_threshold"] == 100 * 1024 * 1024
assert call_kwargs["part_size"] == 8 * 1024 * 1024
assert call_kwargs["num_threads"] == 3
# 验证返回 URL
assert url == "https://test-bucket.oss-cn-hangzhou.aliyuncs.com/test/large.mp4"
finally:
large_file.unlink()
# ── upload_to_oss 超时测试 ────────────────────────────────────────────────────
class TestUploadToOSSTimeout:
"""测试 upload_to_oss() 超时保护."""
def test_upload_timeout_returns_none(self):
"""上传超过总超时时返回 None."""
from video_processing.oss_helpers import upload_to_oss
small_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
small_file.write(b"x" * 1024) # 1KB
small_file.close()
file_path = Path(small_file.name)
def slow_upload(*args, **kwargs):
"""模拟慢速上传,超过超时时间."""
time.sleep(2)
mock_bucket = MagicMock()
mock_bucket.put_object_from_file.side_effect = slow_upload
try:
with (
patch.dict(
os.environ,
{
"OSS_ACCESS_KEY_ID": "test-key",
"OSS_ACCESS_KEY_SECRET": "test-secret",
"OSS_ENDPOINT": "oss-cn-hangzhou.aliyuncs.com",
"OSS_BUCKET_NAME": "test-bucket",
},
),
patch("video_processing.oss_helpers.oss2.Auth"),
patch("video_processing.oss_helpers.oss2.Bucket", return_value=mock_bucket),
patch("video_processing.oss_helpers.OSS_UPLOAD_TOTAL_TIMEOUT", 1), # 1秒超时
):
url = upload_to_oss(file_path, "test/slow.mp4")
# 超时应返回 None
assert url is None, "上传超时应返回 None"
finally:
file_path.unlink()
def test_upload_exception_returns_none(self):
"""上传异常时返回 None."""
from video_processing.oss_helpers import upload_to_oss
small_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
small_file.write(b"x" * 1024)
small_file.close()
file_path = Path(small_file.name)
mock_bucket = MagicMock()
mock_bucket.put_object_from_file.side_effect = RuntimeError("Network error")
try:
with (
patch.dict(
os.environ,
{
"OSS_ACCESS_KEY_ID": "test-key",
"OSS_ACCESS_KEY_SECRET": "test-secret",
"OSS_ENDPOINT": "oss-cn-hangzhou.aliyuncs.com",
"OSS_BUCKET_NAME": "test-bucket",
},
),
patch("video_processing.oss_helpers.oss2.Auth"),
patch("video_processing.oss_helpers.oss2.Bucket", return_value=mock_bucket),
):
url = upload_to_oss(file_path, "test/error.mp4")
assert url is None, "上传异常应返回 None"
finally:
file_path.unlink()
def test_upload_no_bucket_returns_none(self):
"""OSS 未配置时返回 None."""
from video_processing.oss_helpers import upload_to_oss
small_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
small_file.write(b"x" * 1024)
small_file.close()
file_path = Path(small_file.name)
try:
with patch.dict(os.environ, {}, clear=True):
url = upload_to_oss(file_path, "test/noconfig.mp4")
assert url is None
finally:
file_path.unlink()