2020.03.31
Starting around December 2019, some jobs that use near the f_node memory limit (240G) began to run out of enough memory to run the system, causing the node to go down frequently.
As a countermeasure, from April 2020, the upper limit of the amount of memory for f_node by the batch job scheduler has been changed from 240G to 235G to ensure that the system has the necessary memory to operate.
Please refer to the following page for the current specifications of each resource type.
About specification of batch job scheduler: