Derecho compute nodes went down Saturday, April 12, due to an issue with servicing and distributing their filesystem data. Systems engineers quickly triaged the issue and returned Derecho to service several hours later.
All jobs running on Derecho at the time - approximately 10AM MDT - failed. The disruption temporarily interrupted file system activities, which also caused running Casper jobs to fail. For both systems, jobs that were queued at the time of the disruption remained so. For any additionally questions regarding core hour refunds please reach out through
https://rchelp.ucar.edu