1 回答
- 最新
- 投票最多
- 评论最多
0
I've resolved the issue as I suggested previously.
- Platoon or stagger upload requests into arbitrary batch sizes and execute batches in sequence
- Wait for all current batch responses before creating the subsequent batch
- Repeat for all batches
I've requested a documentation modification to indicate the following:
- A single UploadPartCommand must be received by the server within 15 minutes (900 seconds) of its creation
- A single UploadPartCommand that is not received by the server within 15 minutes of its creation will fail with a misleading "RequestTimeTooSkewed" error.
- "RequestTimeTooSkewed" error content, and documentation, and customer support all imply or explicitly state that the error is caused by incorrect client system time. In fact, the problem in this case is that the UploadPartCommand was merely created 15 minutes or more before it was received by the server.
This (#3) can happen as a matter of course when processing very large files. It requires only seconds to generate thousands of UploadPartCommand objects, each of will will have effectively the same creation time. Depending on bandwidth, part size, and part count, it can easily take more than 15 minutes to process these requests. Simultaneous UploadPartCommand requests must be platooned into manageable batches that can be entirely resolved within 15 minutes, lest subsequent, simultaneously created and queued requests fail.
已回答 3 年前
相关内容
- AWS 官方已更新 10 个月前
- AWS 官方已更新 2 年前
- AWS 官方已更新 1 年前