Certain requests run into timeout

[0x7f]

Describe the bug

When hosting min.io (self-hosted S3) via loophole, I can upload large files (multiple 100M or multiple GB) via s3cmd without any issues, but when uploading them via aws-cli, I get timeouts and the upload fails.

To Reproduce Steps to reproduce the behavior:

1. Host min.io e.g. via docker: docker run -v "$HOME/.minio:/data -p 9000:9000 -e "MINIO_ACCESS_KEY=AKIAIOSFODNN7EXAMPLE" -e "MINIO_SECRET_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY" minio/minio server /data
2. Start site via loophole 9000
3. Create S3 bucket via aws-cli: AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY aws s3 mb --endpoint-url https://${LOOPHOLE_SITE}.loophole.site s3://test
4. Check successful upload via s3cmd: s3cmd --access_key=AKIAIOSFODNN7EXAMPLE --secret_key=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY --continue-put --host ${LOOPHOLE_SITE}.loophole.site --host-bucket '${LOOPHOLE_SITE}.loophole.site' put SomeLargeFile.zip s3://test/SomeLargeFile.zip
5. Run into errors when uploading via aws cli: AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY aws s3 cp --endpoint-url https://${LOOPHOLE_SITE}.loophole.site/ SomeLargeFile.zip s3://test/SomeOtherLargeFile.zip

First, download speed goes down from ~1.4M/sec to ~500K/sec and progress hangs every now and then. After some time of no progress, it fails with this message:

upload failed: SomeLargeFile.zip to s3://test/SomeOtherLargeFile.zip Connection was closed before we received a valid response from endpoint URL: "https://${LOOPHOLE_SITE}.loophole.site/max/SomeOtherLargeFile.zip?uploadId=2992f2aa-d82b-4874-bc74-3064a5ae79c5&partNumber=2".

Expected behavior

Upload via s3cmd and aws-cli both work.

Screenshots N/A

Environment

• OS: macOS
• Version: loophole version 1.0.0-beta.7 (95033a34197863fe1074bf148758b30d16477fb6)

Additional context The only difference I can see on a low level is that s3cmd is uploading a single part at a time only and aws-cli is uploading multiple parts in parallel. Maybe this helps. Also somehow suspect the handling of connection close to be the issue.