Due to a configuration error, the scheduler daemon on some compute nodes has timed out, resulting in the nodes being considered as “down” by the scheduler controller. Jobs may have been re-queued as a result.

Things should be coming back to normal now, we apologize for the inconvenience.