KB013: CephRDS Onboarding & Access Guide

Scope: UCR Researchers, External Collaborators, Project Polaris Migrants Audience: All Users Last Updated: Feb 23, 2026

1. Learn More: What is CephRDS?

Ceph Research Data Service (CephRDS) is UCR’s premier on-premise object storage cluster, funded by the NSF CC* program. It provides 2 PB of highly resilient (6+3 Erasure Coded), S3-compatible storage designed specifically for active research data and large-scale computational workflows.

Benefits:

High Performance: Direct connection to the campus Science DMZ and HPCC for rapid data transfer.
Resiliency: Can survive multiple simultaneous drive or node failures without data loss.
S3 Compatibility: Works natively with standard tools like rclone, Python (boto3), and visual clients (Cyberduck, Mountain Duck).
Sustainable Pricing: Built-in hardware refresh cycles via a $40/TB/Year rental rate or a $200/Usable TB one-time permanent purchase.

2. Get Started & Get an Account

We operate a Split Allocation model to provide the best experience and ensure security depending on your affiliation.

A. UCR Faculty & Internal Labs

If you are a UCR PI needing storage for your lab, or migrating from an older system:

Request Access: Email research-computing@ucr.edu with your requested storage quota (in TB) and your grant Chart of Accounts (COA) for billing (if exceeding the initial allocation).
Provisioning: Our team will manually provision your account and establish your strict user quota.
Credential Delivery: You will receive your S3 Access Key and Secret Key via a secure, one-time link.

B. External Collaborators on UCR Projects

If you are an external collaborator (e.g., from UCSD, or another institution) working with a UCR lab on data hosted in a UCR bucket, we do not use an external federation portal.

PI Approval: The UCR Principal Investigator (PI) must email research-computing@ucr.edu or submit a BearHelp ticket explicitly requesting access for you.
Provisioning: Our team will manually generate a unique set of Access/Secret keys scoped strictly to the UCR PI’s bucket.
Delivery: The credentials will be securely delivered to the external collaborator.

3. Get Your Token (S3 Keys)

Unlike standard campus services, CephRDS uses the S3 API Protocol. This means you do not use your NetID and password to log in interactively. Instead, you use an Access Key and a Secret Key.

Security Warning: Treat your Secret Key like a highly sensitive password. Never commit it to GitHub or share it in plaintext.

4. Get Connected & Transfer Data

We recommend different tools based on your operating system and technical comfort level.

Method A: Visual Interface (Windows / Mac)

For a drag-and-drop experience similar to Google Drive or Dropbox, we recommend Cyberduck (Free) or Mountain Duck (Paid, mounts as a local drive).

Note: We do not recommend or support FileZilla for CephRDS as it does not natively support the modern S3 protocols required for our deployment.

Setup Cyberduck:

Download and install Cyberduck.
Click Open Connection.
Select Amazon S3 from the dropdown menu.
Server: s3.ucr.edu (Note: exact endpoint pending final DNS configuration)
Port: 443
Access Key ID: Paste your provided Access Key.
Secret Access Key: Paste your provided Secret Key.
Click Connect.

Method B: Command Line & HPCC (Linux)

For moving massive datasets (terabytes) or running scheduled jobs on the HPCC, rclone is the standard. It can saturate 100Gbps network links through parallel transfers.

Setup rclone:

Run rclone config
Press n for New remote.
Name it cephrds.
Choose s3 (Amazon S3 Compliant Storage Providers).
Choose Ceph as the provider.
Enter false to enter credentials manually.
Enter your Access Key ID.
Enter your Secret Access Key.
Endpoint: https://s3.ucr.edu

Example Transfer Command: Copy a local folder to a bucket on CephRDS:

rclone copy /local/data/my_dataset cephrds:my_project_bucket/ --progress --transfers 8

Method C: Python Scripting (`boto3`)

If you are writing data analysis pipelines in Python:

import boto3

# Initialize the S3 client
s3 = boto3.client('s3',
    endpoint_url='https://s3.ucr.edu',
    aws_access_key_id='YOUR_ACCESS_KEY',
    aws_secret_access_key='YOUR_SECRET_KEY'
)

# List all buckets
response = s3.list_buckets()
print(response['Buckets'])

5. Troubleshooting & FAQ

Q: I got a “403 Forbidden” error when trying to upload. A: You have likely hit your assigned storage or object quota. Please contact support to request a quota increase.

Q: I lost my Secret Key. A: Secret Keys cannot be recovered, only regenerated. Contact support to generate a new key pair. You will need to update all your scripts and clients.