Your own bucket · Docs

By default a server stores files in CloudFile storage, which counts towards your plan. You can instead connect your own S3-compatible bucket. SFTP and FTP uploads then go straight into it, and the storage is billed by your provider, not by us. This works on every plan.

Connect a bucket

Go to Storage and click Add bucket.
Pick your provider. We fill in the endpoint and region format for you.
Enter the bucket name, access key ID and secret access key.
Click Test connection. We write a small .cloudfile-check file, read it back, list the bucket and delete the file again.
Save.

Then open a server, go to Settings → Storage, pick the bucket and save. You can also pick it when you create a server.

Each provider below has a step-by-step setup guide. Any other service that speaks the S3 API works too: pick Other S3-compatible.

Choosing a provider

Any of these works the same way in CloudFile. The differences are in pricing and where the data lives:

Provider	Good for	Notes
Amazon S3	Teams already on AWS	Most regions; egress is billed per GB
Cloudflare R2	Servers with heavy downloads	No egress fees; optional EU jurisdiction
Wasabi	Long-term archives	No egress fees while monthly downloads stay within stored volume; 90-day minimum storage on pay-as-you-go
Backblaze B2	Low-cost storage	Free egress up to 3× stored data; one region per account
DigitalOcean Spaces	Teams already on DigitalOcean	$5/month includes 250 GiB storage and 1 TiB outbound transfer
Hetzner Object Storage	EU data residency	Germany and Finland only; monthly price includes a storage and egress quota
Google Cloud Storage	Teams already on Google Cloud	Egress is billed per GB; soft delete keeps deleted files (billed) for 7 days by default
Scaleway	EU data residency	Paris, Amsterdam, Warsaw and Milan
OVHcloud	EU data residency, global reach	Regions in Europe, Canada, Singapore, Australia and India
IDrive e2	Low-cost storage	Regions in the US, Canada, Europe and Asia
MinIO	Files on your own hardware	You run the server; it must be reachable over public `https`

Where files go

Each SFTP user gets their own folder in the bucket:

<bucket>/<folder>/<username>/<path>

The folder is optional and set per server in Settings → Storage. Leave it empty to put user folders at the root of the bucket. Several servers can share one bucket if each uses a different folder.

Permissions

The access key needs these permissions on the bucket:

Permission	Used for
`s3:ListBucket`	Listing folders
`s3:GetObject`	Downloads and download links
`s3:PutObject`	Uploads, renames and new folders
`s3:DeleteObject`	Deletes and renames
`s3:AbortMultipartUpload`	Cleaning up interrupted large uploads

Limit the key to this one bucket. We encrypt the secret key before storing it and never show it again.

Switching storage

You can move a server between CloudFile storage and your buckets at any time. Files are not copied. After the switch, users see what is in the new location. To keep existing files, copy them first, for example with rclone, into <folder>/<username>/.

Deleting a server or a user never deletes files in your bucket.

Quotas and webhooks

User quotas still work on your own bucket and don’t use your plan’s storage. Quota usage is measured by scanning the user’s folder, so very large folders take a little longer to update.

Webhook payloads include a download_url. On your own bucket it’s a presigned link to your bucket, valid for one hour.

Requirements

The endpoint must use https and be reachable from the internet.
The bucket must already exist. We never create or delete buckets.
Path-style URLs (https://endpoint/bucket/key) are used for Cloudflare R2, Google Cloud Storage, IDrive e2, MinIO and Other S3-compatible, and are needed by most self-hosted services such as Ceph. Virtual-hosted URLs (https://bucket.endpoint/key) are used for the other providers. The provider presets pick the right one.