As previously announced, CISL staff will be performing necessary facility maintenance at the NCAR-Wyoming Supercomputing Center (NWSC) beginning on June 2 to replace critical power and cooling systems which have reached their end service life. Because of the nature of this work, it will be necessary to shut down all Derecho compute nodes, the Quasar tape archive, and a portion of the Casper compute nodes. We aim to restore service on June 6, but the outage may extend to June 9 if required to complete the planned work.

CISL intends to keep Derecho and Casper login nodes, a subset of Casper high-throughput computing nodes, and the GLADE file systems (home, Campaign Storage, and scratch) online throughout the outage using temporary, augmented cooling. JupyterHub access to Casper is also anticipated. However, additional hardware may need to be shut down to maintain safe operations if thermal conditions dictate. Additionally, a portion of the available Casper compute resources will be used by a workshop throughout the week.

As a result, access to compute nodes on Casper will be very limited (and GPU resources will be unavailable), so we suggest only submitting time-sensitive work during the outage. We understand extended downtimes are difficult for the user community and have made every effort to lessen the impact of this critically important maintenance activity given all constraints.