As part of the ongoing effort to upgrade data center capacity for future computing needs, maintenance is scheduled from Wednesday, April 22 at 6 AM through Thursday, April 23 at 5 PM. This will affect all RCAC systems and the Research Network.

During this window, all RCAC systems and research network services will be unavailable, including:

• All Computing Clusters including Bell, Negishi, Gautschi, Scholar, Rowdy, Gilbreth, Hammer, Anvil.
• All Data Storage Systems including Data Depot, Fortress, Anvil Ceph storage, scratch and home storage on clusters, Research Network and ScienceDMZ
• Gateway services including Hubzero, GenAI Studio, Anvil GPT
• www.rcac.purdue.edu
• Geddes
• Globus

How does this maintenance impact you?

• Any Slurm jobs requesting a walltime that would extend past the start of the maintenance will not start and will remain in the queue until after maintenance is complete.
• All active sessions and jobs running on affected systems will be preempted at the start of the outage any queued jobs will not begin until services are restored.
• Access to login nodes, storage systems, and web portals will be unavailable throughout the downtime.
• Automated data workflows that rely on affected systems (e.g., rsync, data pipelines, archive processes) will not function until systems are back online.
• Globus transfers may time out during the maintenance window.

To prepare for this maintenance, we suggest to:

• Download any needed data or scripts before 6:00 AM on April 22.
• Prepare instrumentation devices for the Data Depot to be unavailable and

We appreciate your understanding and cooperation as we complete these necessary upgrades to improve reliability and performance across RCAC infrastructure. For assistance, questions, or concerns contact rcac-help@purdue.edu.

Best regards, Rosen Center for Advanced Computing (RCAC) / Purdue IT

To prevent further impact, job submissions to clusters connected to Depot have been temporarily paused while the team works restore normal operation.

Impact: Users may experience job delays or failures when writing to the Data Depot.

Current Status: Our engineering team is actively addressing the capacity issue.

Next Update: We will provide an update once the service is restored or we have more information to share.

Impact to you Attempts to read files that are on the affected storage may result in error messages

Our IT team is actively working with the vendor to restore service as quickly as possible. We will send an update as soon as more information is available.

We will provide updates as the situation develops and once service is fully restored.

We appreciate your patience and understanding.

Impact To You: At this time, you may notice latency or brief disruptions when accessing certain on-campus or external resources, especially during peak usage periods.

We appreciate your patience while we work to fully resolve the underlying problem and restore normal network performance. We will provide an update by 5:00PM EST today or sooner.

System monitoring has identified an issue where heavy job activity was overloading the Data Depot filesystem used by the clusters and Globus.

You may see the following impacts:

• Globus transfers to and from Depot collections may fail, stall, or run much more slowly than usual.
• Globus logins may be slow or occasionally fail when accessing Depot endpoints.
• Jobs on RCAC clusters that read from or write to Depot may experience slow file access, delayed directory listings, or timeouts.

Our engineers are investigating the high load from a large number of concurrent jobs and are working to reduce the impact on Depot, Globus, and cluster workloads. Existing jobs will continue to run, but any that are heavily Depot‑I/O‑bound may run more slowly or see I/O errors until performance improves. We will provide another update by 5:00PM EST or sooner if the issue is resolved.

We will provide an update by 2:00PM EST today.

Research Computing staff members will monitor the status of all computing and data resources in an effort to ensure continuous availability.

Research Computing staff members will monitor the ticketing system throughout the holiday period and answer critical issues and problems. Non-critical user issues and questions will be addressed beginning January 5th, 2026. There will also be no coffee hour consultations during this break.

Scratch file purging (on community clusters with scratch space) will continue as normal during the break, so be sure to archive your files in scratch storage. This does not apply to Data Depot or home directories -- only scratch storage.

Have a wonderful break, everyone, and we look forward to great things in the new year!

We will provide an update by 9pm.

We will provide an update once the maintenance has been completed.

Powered by IBM Storage Scale filesystem via partnership with Dell Technologies and DataCore Nexus, the Data Depot uses an enterprise-class GPFS storage solution with a total capacity of 10 petabytes (PB). This storage is redundant and reliable, features regular snapshots, and is globally available on all RCAC systems. The Data Depot is a non-purged space suitable for tasks such as sharing data, editing files, developing and building software, and many other uses.

The Data Depot facilitates joint work on shared files across research groups, avoiding the need for numerous copies of datasets across individuals' home or scratch directories. It is an ideal place to store group applications, tools, scripts, and documents. With this upgrade, Data Depot’s expanded capabilities will allow researchers to accomplish even more with their data. For example, massive datasets, often required for advanced artificial intelligence (AI) workloads, will be able to be stored and quickly retrieved for immediate user access. This allows researchers to conduct studies of a larger scope and scale than previously possible, while also enabling a faster time-to-science.

Improved Storage and Performance:

The third generation of Data Depot has been upgraded to better support the emerging needs of today’s research environment. The Data Depot now includes:

• 10 PB total capacity, with 2 PB of flash to maximize performance
• Low latency, Remote Direct Memory Access (RDMA) connectivity to HPC systems, to better enable AI workloads
• Total raw performance of 500 GB/sec, a 25x increase in total performance
• A powerful data tiering policy engine
• Cloud bursting and syncing tools available

These enhancements to Data Depot’s capabilities are now stacked on top of the features that over 700 research labs have come to depend on to enable their science. Features like Depot’s persistent (not purged) storage, redundancy, and self-service management of research lab permissions allow a PI to effectively govern and manage their lab’s data.

Data Depot is accessible as a Windows or Mac OS X network drive on personal and lab computers on campus, directly on Community Cluster nodes, or from other universities or labs through Globus. No matter where a research group works, Depot provides ubiquitous access to the lab’s data.

Higher performance, lower price

Along with changes to its storage capabilities, Data Depot v3 now offers more affordable pricing. The new price for Data Depot is $47 per TB per year, down from $70 per TB/year. Thanks to Purdue’s purchasing power and innovations in storage hardware, the third generation of Depot has resulted in one that is bigger, faster, and cheaper for Purdue researchers to use. Data Depot is available to any Purdue research group as an annual purchase in increments of 1TB. For groups who are unsure or would like to experience before buying, RCAC offers a 100 GB trial space, free of charge. Note: Participation in the Community Cluster program is not required. To purchase capacity on Data Depot, please visit RCAC’s dedicated Purchase Page.

A complete ecosystem for research data

The Data Depot filesystem is part of a whole ecosystem in support of data for open science, which include:

• Depot object storage: Depot object storage is a high-capacity, fast, reliable, and secure object storage service. Funded by an NSF CC* award, Depot object is a high-performance Ceph storage solution with an initial total capacity of over 4 PB. This storage is redundant and reliable, with APIs accessible from any Purdue network.
• Fortress Tape archive: The Fortress system is a large, long-term, multi-tiered file caching and storage system utilizing both online disk and robotic tape library, ideal for near-line caching of raw data, or archiving completed projects.

Depot and Fortress are part of an integrated high-performance computing and storage environment, with Globus-based links to the Purdue University Research Repository (PURR), and easy connectivity to experimental instrumentation facilities. RCAC also provides equivalent offerings for export-controlled data (via the Weber cluster), and sensitive and restricted data, including life sciences (via the Rossmann cluster, which is supported by the Protected Data Filesystem (PDFS) and the Protected Data Archive (PDA)).

“RCAC has always had the mandate to provide Purdue with world-class cyberinfrastructure at the highest proven value, which means powerful solutions at attractive prices,” says Preston Smith, executive director of RCAC. “With this newest generation of the Data Depot filesystem entering production, we look forward to it continuing to provide a flexible and accessible solution to ensure that Purdue researchers’ scientific data are in a resource right where it can be processed or simulated. Depot V3’s speed and accelerated connectivity will further augment the capabilities around artificial intelligence, and the Pixstor software by DataCore Nexus will allow for more effective management of our researchers’ data lifecycle.”

Researchers who are unsure which data storage solutions are right for them should utilize RCAC’s Storage Solutions Finder, an interactive tool to help Purdue researchers compare and select a variety of available storage options based on their usage needs and data security constraints.

RCAC operates all centrally-maintained research computing resources at Purdue University, providing access to leading-edge computational and data storage systems as well as expertise and support to Purdue faculty, staff, and student researchers. To learn more about HPC and how RCAC can help you, please visit: https://www.rcac.purdue.edu/

Written by: Jonathan Poole, poole43@purdue.edu

The Data Depot file system has returned to full service and scheduling has resumed on all clusters.

The Data Depot storage system began experience issues starting around 9am EDT this morning. Engineers are currently diagnosing the issue and are working to identify a fix. Job scheduling has been paused while this issue is being addressed.

We will provide an update by 12pm (noon).

Data Depot functionality has been restored.

The Data Depot file system began experiencing issues with writes around 2:30pm EDT. The data migration process currently ongoing from Data Depot 2 to Data Depot 3 ran into an unexpected problem. Engineers have identified the problem and are correcting it. Users may have seen "no space left on device" for approximately 30 minutes. Job scheduling has been paused while this issue is being addressed.

We will provide an update by 5 PM.

We know how important timely access is to your research and truly appreciate your patience as this gets sorted out. If you have an urgent need or specific concerns, please don’t hesitate to contact us at rcac-help@purdue.edu.

Thanks again for your understanding.

We are actively monitoring the situation and working to resolve the disruptions as quickly as possible. During this time, you may encounter intermittent connectivity problems or degraded performance.

We appreciate your patience and understanding. If you encounter problems that require immediate attention, please contact rcac-help@purdue.edu. For Anvil users, please submit a ticket through ACCESS Help Desk at https://support.access-ci.org/help-ticket.

Engineers are currently diagnosing the issue.

We will provide an update by 8pm.

Engineers are currently diagnosing the issue.

Research Computing staff members will monitor the status of all computing and data resources in an effort to ensure continuous availability.

Research Computing staff members will monitor the ticketing system throughout the holiday period and answer critical issues and problems. Non-critical user issues and questions will be addressed beginning Thursday, January 2nd, 2025. There will also be no coffee hour consultations during this break.

Scratch file purging (on community clusters with scratch space) will continue as normal during the break, so be sure to archive your files in scratch storage or check for any upcoming purge warning emails during the break. This does not apply to Data Depot or home directories -- only scratch storage.

Have a wonderful break, everyone, and we look forward to great things in the new year!

However, due to reduced staff availability during this period we will be canceling our regularly scheduled Coffee-Hour consultations session for that week. This includes both the Monday in-person consulting hours at the Convergence Center as well as the Tuesday virtual sessions.

We will provide an update by 5PM today.