October MGHPCC Data Center Shutdown
Scheduled Maintenance Report for IT Services - Northeastern University
Completed
The scheduled maintenance has been completed.
Posted Oct 22, 2020 - 17:00 EDT
In progress
Scheduled maintenance is currently in progress. We will provide updates as necessary.
Posted Oct 19, 2020 - 05:30 EDT
Scheduled
The Massachusetts Green High Performance Computing Center (MGHPCC) will be conducting its annual shutdown for maintenance work starting October 19th. Northeastern’s Discovery cluster is physically located at MGHPCC and is included in this shutdown, along with the high performance cluster systems of Harvard, Boston University, Tufts, and the University of Massachusetts at Amherst. Discovery will be powered down starting at 5:30 a.m. on October 19th. The shutdown is scheduled to last for two days (October 20 and 21), and we are currently anticipating that the Discovery cluster will be powered back on by 6 p.m. on October 22, given that all work is completed in a timely manner.
This routine maintenance is normally performed in July, but due to the COVID-19 pandemic and resulting lockdown in the state of Massachusetts, this maintenance window was moved to October. In addition, out of an abundance of caution, this year’s maintenance window includes a staggered schedule for all participating universities. This is to help facilitate social distancing for the system administrators that will be on site for this time period. The result of this staggered schedule is a slightly extended outage period from previous MGHPCC shutdowns.
All systems are affected by the shutdown, including all compute nodes, storage (including research and archive storage), and the Open onDemand web portal. Please be aware that any running jobs will be terminated on October 19th as the system is powered down.
If you have any questions or concerns, please reach out to us at rchelp@northeastern.edu.
The Research Computing Team
Posted Sep 21, 2020 - 12:58 EDT
This scheduled maintenance affected: Research Computing (Discovery Cluster Compute, Schrodinger / Maestro).