aws / aws/sagemaker-python-sdk

Cannot upload local file for submit_py_files in V3

未关闭
#6,252 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
主要语言
Python
星标
2.3k
派生
1.3k
平均合并
1 天 22 小时
30 天内合并 PR
35

描述

**PySDK Version**

- [ ] PySDK V2 (2.x)
- [x] PySDK V3 (3.x)

**Describe the bug**

The PySparkProcessor `run` no longer accepts local files due to the change to `ProcessingInput` interface - e.g. regression introduced in 3203e49

**To reproduce**

Insert a local file path into the list passed as `submit_py_files` in `sagemaker.core.spark.processing.PySparkProcessor.run()`.

(Let me know if you need a more concrete example).

**Expected behavior**

Successfully starts the job.

**Screenshots or logs**

PyDantic reports the following errors:

```
ValidationError: 2 validation errors for ProcessingInput
source
Extra inputs are not permitted [type=extra_forbidden, input_value='s3://sagemaker-.../input/py-files', input_type=str]
For further information visit https://errors.pydantic.dev/2.13/v/extra_forbidden
destination
Extra inputs are not permitted [type=extra_forbidden, input_value='/opt/ml/processing/input/py-files', input_type=str]
For further information visit https://errors.pydantic.dev/2.13/v/extra_forbidden
```

The cause of the error is [here](https://github.com/aws/sagemaker-python-sdk/blob/8e7485a1ed25eb17c70f2323ab6ed6c9695d7024/sagemaker-core/src/sagemaker/core/spark/processing.py#L508-L512):

```py
input_channel = ProcessingInput(
source=input_channel_s3_uri,
destination=f"{self._conf_container_base_path}{input_channel_name}",
input_name=input_channel_name,
)
```

**System information**

- AWS SageMaker Studio 4.4.3
- sagemaker-core: 2.20
- Python 3.12.14
- uname: Linux default 6.12.103-127.188.amzn2023.x86_64 SMP PREEMPT_DYNAMIC Tue Aug 25 15:42:53 UTC 2026 x86_64 x86_64 x86_64 GNU/Linux

**Additional context**

贡献指南

打开贡献指南

调研方向

从 sagemaker/core/spark/processing.py 中 PySparkProcessor.run() 的代码(约第 508-512 行)开始,跟踪 submit_py_files 中的本地路径如何转换为 ProcessingInput。完成标准是接受本地文件路径,并且处理作业能够成功启动,同时不会出现所报告的 Pydantic 验证错误。

由索引模型根据 Issue 内容生成。

评估

技术栈
aws, python, spark
领域
backend-api-design, machine-learning
Issue 类型
缺陷
难度
3/5
预计耗时
1-2 天
活跃度
活跃
描述清晰度
描述清楚
新手友好度
78/100

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。