Mount a Bucket as a Filesystem
A bucket can be mounted as a local filesystem, letting tools that expect POSIX paths read and write objects without any code changes. This is useful for migrations, backup tools, build pipelines, or any utility that doesn’t speak S3 natively.
Object Storage isn’t a real POSIX filesystem. Mounted access works well for sequential reads, batch writes, and general file workflows — but avoid it for random small writes or anything that depends on file locking.
Before you begin
- A bucket (Create a bucket)
- An API credential with at least Object Read & Write scope on the target bucket
- The S3 endpoint for your organization:
https://{workspaceId}.blob.ibeestorage.com
Pick a tool
s3fs (Linux / macOS)
Install:
Save your credentials:
Mount the bucket:
Now /mnt/my-bucket reads and writes against the bucket.
Unmount:
Auto-mount at boot (Linux)
Add to /etc/fstab:
rclone
Install: rclone.org/install
Configure a remote:
Mount:
Or sync without mounting:
Performance notes
- Sequential is fast, random is slow. Mounted filesystems shine for whole-file reads and writes; small in-place edits trigger full re-uploads.
- Cache aggressively. Tools like
rcloneands3fshave local cache options — use them. - Don’t use as a database backing store. No file locking, no atomic appends.
Common pitfalls
- Symlinks are not real symlinks — they become regular files containing the link path.
- File modification time comes from object metadata; tools that rely on
mtimemay behave unexpectedly. duis slow because it has to list every object.
