Two nodes crashed yesterday (Apr. 16) while running jobs dispatched by the same user. While we are still investigating the cause, we believe the most likely cause was the nodes running out of memory. As a precaution, we are reducing the amount of usable memory as follows:

Nodename Old Max. Mem. (GiB) New Max. Mem. (GiB)
scrp-node-[4,5,8] 62 60
scrp-node-12 250 247
scrp-node-[1-3,6,9] 500 497
scrp-node-11 752 747
scrp-node-[7,10] 991 987