Dear CRCD Users,
To preserve login node performance, we have enforced per-user resource limits of 1 CPU core and 8 GB of memory.
These limits apply collectively — regardless of how many processes a user runs, their combined usage will not exceed these thresholds.
Important: When the login node's resource limits are reached, users may be unable to log in until utilization falls below the threshold.
We strongly remind all users to avoid running jobs directly on the login node. Even within these limits, compute- or I/O-intensive workloads can degrade performance for everyone. Processes that violate this policy will be terminated without notice.
Running Work on the Cluster
For interactive sessions requiring significant resources, please request a dedicated compute node via an interactive job instead:
bash
srun -M smp --export=ALL --nodes=1 --time=24:00:00 --mem=16g --ntasks-per-node=8 --pty bash
Keeping Sessions Alive
If you need to keep a session running after disconnecting or to prevent logout due to keyboard inactivity, please use nohup or tmux rather than running long processes on the login node. See our user manual for guidance on using nohup and tmux on the cluster.
VS Code Users
For VS Code support, please see the three options described in our manual.
Sincerely,
The CRCD Team