Skip to content

Using the RustFS Storage Provider

RustFS is an S3-compatible object storage system that OSCAR can use instead of MinIO as its storage provider. As with MinIO, the storage provider is responsible for triggering services: when a service is configured to watch an input folder, any new data added to that folder fires an event that starts a job of the associated service.

This document describes how OSCAR integrates with RustFS, how to configure it, and what differs from the MinIO storage provider.

What is RustFS?

RustFS is an object storage system that is compatible with the S3 API for standard bucket and object operations. OSCAR treats RustFS as a drop-in replacement for MinIO: the same policy creation, assignment, update and removal flow is used, but routed through the RustFS-native IAM API.

Choosing RustFS: configuration

The storage backend is selected at OSCAR startup with the OBJECT_STORAGE_TYPE environment variable. It accepts two values:

Value Meaning
minio Use MinIO (default).
rustfs Use RustFS. 
OBJECT_STORAGE_TYPE=rustfs

When deploying OSCAR with Helm, the same setting is forwarded through the minIO.object_storage_type chart value. The rest of the object storage settings (endpoint, credentials, TLS verification) keep the minIO.* prefix and behave exactly as they do for MinIO:

helm install --namespace=oscar oscar grycap/oscar \
  --set minIO.object_storage_type=rustfs \
  --set minIO.endpoint=http://rustfs-svc.rustfs:9000 \
  --set minIO.TLSVerify=false \
  --set minIO.accessKey=rustfs --set minIO.secretKey=<RUSTFS_PASSWORD>

Because RustFS supports the S3 API, the minIO.* endpoint and credential settings point to the RustFS S3 endpoint, and the OSCAR web interface (including the buckets and files sections) works in the same way as with MinIO.

How OSCAR interacts with RustFS

OSCAR distinguishes the backend at runtime through the configured OBJECT_STORAGE_TYPE value. All IAM and notification operations are then performed against the RustFS native admin API at /rustfs/admin/v3, with every request signed using AWS Signature Version 4 (service s3).

The main routes used by OSCAR are:

Route Method Purpose
/add-user PUT Create an OSCAR user.
/update-group-members PUT Manage group membership (RustFS JSON format).
/add-canned-policy PUT Create a policy document.
/set-user-or-group-policy PUT Attach a policy to a user or group.
/target/notify_webhook/<service> PUT Register a per-service webhook target.
/target/notify_webhook/<service>/reset DELETE Remove a per-service webhook target.
/module-switches GET/PUT Read and enable the notification module.

Bucket management and visibility

Metadata stored as bucket tags

RustFS uses group-based policies to model bucket visibility. Also, RustFS and OSCAR use metadata to store information:

Tag Meaning
owner User ID of the bucket owner.
owner_name Username of the bucket owner.
visibility public, restricted or private.
allowed_users Space-separated list of users allowed to access a restricted bucket.
storage_quota Per-bucket storage limit (when a quota is set).

Bucket quotas

Per-bucket storage limits configured through the /system/quotas API (storage_per_bucket) are enforced with the MinIO-compatible bucket quota admin API (set-bucket-quota), which RustFS implements.

Deploying RustFS

Local deployment

The local deployment script installs RustFS automatically when it is selected as the storage backend:

bash oscar/deploy/kind-deploy.sh --storage=rustfs

This:

  • Installs the rustfs/rustfs Helm chart in the rustfs namespace, using the standalone mode (no erasure-coded distributed layout).
  • Creates the rustfs-svc service in the rustfs namespace; OSCAR is pointed at http://rustfs-svc.rustfs:<port>.
  • Exposes the S3 endpoint and the RustFS console on the same node ports that MinIO uses (30300 and 30301 by default).
  • Sets RUSTFS_OUTBOUND_ALLOW_ORIGINS=http://oscar.oscar:8080 so the OSCAR job endpoint is a valid webhook target.
  • Configures OSCAR with OBJECT_STORAGE_TYPE=rustfs through the minIO.object_storage_type chart value.

The --minio-quotas option is not compatible with RustFS: it prepares an erasure-coded distributed MinIO layout (one replica, four drives) which is a MinIO-only prerequisite. Use it only when MinIO is selected.

Accessing RustFS buckets

Because RustFS is S3-compatible, all the access methods documented for MinIO apply to RustFS as well:

  • OSCAR Dashboard: the buckets section lists the buckets visible to the authenticated user and allows uploading and downloading files, exactly as with MinIO.
  • oscar-cli: the get-file, list-files and put-file commands operate on the configured storage provider regardless of the backend.
  • Any S3 client: standard S3 tools (for example mc) can be pointed at the RustFS S3 endpoint with the user credentials to manage files inside buckets.