Buckets & objects
Use a bucket to organize a project’s objects. Access them by bucket name and case-sensitive key. The Dashboard endpoints manage keys, buckets, sessions, and usage; the S3 endpoint handles signed object requests directly.
Buckets
Create a bucket before uploading objects. The service validates S3-style names and keeps names unique within the Storage tenant. Bucket settings include versioning and image variant retention. A bucket must be emptied before deletion; deleting a bucket is not a substitute for an object lifecycle policy.
Object keys and metadata
Keys are case-sensitive. A slash is part of the key, so images/logo.png and Images/logo.png are different objects. Use prefixes to list a folder-like subset. Set Content-Type on upload so downloads and media processing identify the object correctly. The service records size, ETag, SHA-256, content type, and creation time.
Upload and download
| Method | Use it when | Credential |
|---|---|---|
S3 PUT/GET | A trusted server or SDK reads and writes files directly. | Scoped S3 access key with SigV4 |
| Dashboard upload session | A browser or client needs a temporary upload URL. | Dashboard authorization to create session; temporary URL for the bytes |
| S3 multipart | A large object should be sent in parts that can be completed or aborted. | Scoped S3 access key with write permission |
The upload-session response provides upload_url, object_key, content_type, and expires_at. PUT the content to that URL before expiry. The matching download endpoint returns a temporary download_url. These URLs are capabilities; share them only with their intended user.
Multipart operations are initiate → upload numbered parts → complete. Abort an unfinished upload to release staged data. Multipart parts count toward usage until completed or cleaned up. S3 supports server-side copy and byte-range GET requests as well.
From a terminal, hubfly storage uses the same upload and download sessions, one request per file of up to 20 GB:
hubfly storage buckets create project-assets
hubfly storage put ./dist project-assets/site/ --recursive # keys keep the relative paths
hubfly storage ls project-assets/site/
hubfly storage get project-assets/site/index.html - # write to standard output
hubfly storage url project-assets/site/index.html # temporary download URL
hubfly storage rm project-assets/site/ --recursive # lists, counts and confirms firstVersioning and deletion
Bucket versioning can be disabled, enabled, or suspended. When enabled, replacement and deletion can leave older versions or delete markers. Plan cleanup and quota around these retained bytes. Deleting a managed asset or a generated rendition can also break URLs already used by an application; review references before removing them.
Check the exact target
Listing and pagination
Object listing accepts a prefix and returns a continuation token when more results are available. Dashboard browsing also accepts a search term, view, and page size. Continue until no next token is returned; do not assume one response contains the entire bucket. The S3 list operation caps a page at 1,000 objects.