Folder disappears after deleting all files contained within it about mountpoint-s3 HOT 5 OPEN

S3 does not support "folders", the objects are being grouped based on the path, to avoid this you'd need to keep at least one object.

Amazon S3 has a flat structure instead of a hierarchy like you would see in a file system.

The console does this by using a shared name prefix for the grouped objects. In other words, the grouped objects have names that begin with a common string. This common string, or shared prefix, is the folder name.
https://docs.aws.amazon.com/AmazonS3/latest/userguide/using-folders.html

from mountpoint-s3.

Hey @revanhernandy. As @mschfh wrote, the objects are being grouped based on path. This is also described in our semantics documentation for reference: https://github.com/awslabs/mountpoint-s3/blob/83fa7c654623cba34b44f99c3b8930b693ebd0b1/doc/SEMANTICS.md#directories

One way to resolve your issue would be to create either an empty file for persisting a directory (similar to .gitkeep files for Git), or to create a directory marker. We don't recommended using directory markers generally since they can have adverse effects with larger workloads, but it may work best for your use case.

A directory marker is an empty object with the key matching the prefix of objects that would be contained within. For some directory "dir1", you could create an empty object dir1/ which will be interpreted by many file-oriented connectors/views as a directory. For example, the S3 Console will create this type of object when creating a directory. Note that directory markers cannot be removed by Mountpoint today, and you would need to use another S3 client to remove them later.

Please let us know if this resolves your issue or if you need any further help with it.

from mountpoint-s3.

Hi @dannycjones thank you for the response. Got it and makes sense.
Currently, we have an Azure blobfuse filesystem and it can be handled by a temporary path but we need to migrate into S3 stuff 😄 it would be very perfect if Mount S3 could do similar.
But, the first workaround is ideal in this case with making a persistence file with a dot(.) would be helpful in case, vsftpd by default would hide the files and directories starting with dot(.)

from mountpoint-s3.

Currently, we have an Azure blobfuse filesystem and it can be handled by a temporary path but we need to migrate into S3 stuff 😄 it would be very perfect if Mount S3 could do similar.

Sorry @revanhernandy, I didn't understand what the ask is. What do you mean by a temporary path?

from mountpoint-s3.

We plan to add the scenario in this issue to a new FAQ/troubleshooting document we're creating, tracking this scenario here: #722.

from mountpoint-s3.