Following the recent downtime, the amount of memory available for jobs on nodes has been tuned to increase system stability. On all nodes except intel18, the amount of memory available to jobs has been reduced by 5GB. A full list of node types and the memory available to jobs can be found in our documentation. This adjustment will help prevent unexpected lockups and reboots of nodes during times of high memory utilization.

Jobs scripts that request the maximum amount of memory available on a particular node type may need to be adjusted following this change. If you have any questions about this change, please contact us.