aws s3 sync fails if there is a directory in the target location with the same name as a file in the source location

[pako-pl]

$ aws s3 sync --delete s3://my-test-bucket/ /backup/my-test-bucket/
download failed: s3://my-test-bucket/item.208 to ../backup/my-test-bucket/item.208 [Errno 21] Is a directory

item.208 is a file in the source S3 bucket and a directory in the target local directory. The situation occured after we moved around some files in S3, including replacing a directory with a file with the same name. It caused our backups to stop running due to "[Errno 21] Is a directory". We would expect "s3 sync" with the "--delete" flag to be able to cope with such a situation.

$ aws --version
aws-cli/1.14.63 Python/2.7.13 Linux/4.9.0-6-amd64 botocore/1.9.16

[kyleknap]

@pako-pl Can you clarify what the expected behavior would be for this scenario? Would it be to?

• Ignore the download of item.208 from S3
• Delete the directory item.208 and replace it with the object item.208 from S3

I think it would be the latter, but it may not be possible to do because the commands are trying to map a key/value store to a file system which has caused other issues like this in the past.

[pako-pl]

The expected behavior would be to delete the item.208 directory from the target and replace it with the item.208 object from S3, as I am running a "sync" with the "--delete" flag.

[arochettesteinberg]

I think I have the same issue here. Any progression? Workaround?

[linture]

same issue here

[stani-o]

This issue happens also when you try to sync a s3 bucket that has equally named files and directories under the same hierarchy level. I assume this happens because the s3 sync download's first the equally named folder locally and then fails when it download's the equally named file.

Can be easily reproduced:

1. Create a file named "test"
2. Create a folder "test"
3. Create a file test2 in the folder test/
4. aws s3 sync s3://bucket/path .