Update 7/7 9AM: DCC maintenance has completed and the DCC has been returned to service. The DCC team is still working to resolve issues on a small number of nodes that did not patch correctly and still remain down.
Update 7/6 8PM: The DCC has been returned to partial service and users can again submit jobs. Remaining nodes will be returned to service throughout the next 12 hours.
Please note: for GPU nodes, CUDA version has been updated to 11.0
Beginning at 9 AM on 7/6/21, the Duke Compute Cluster will be shut down for routine system patching and other operational updates. All running and pending jobs will be cancelled and the login nodes will be unavailable to users. As patching progresses, partial services on the DCC will be restored later in the day on 7/6, with the majority of nodes back in service within 24 hours.
- Duke Compute Cluster (DCC)
What’s going to be done during the outage?
- SLURM software update to the current stable version
- Operating system patching (for GPU hosts, CUDA version will now be 11.0)
- Enhanced filesystem security for group directories
Questions about the outage and the changes should be emailed to email@example.com.