GPU Cluster: XFACT Lab
Status: Completed
Maintained the shared backbone of a 6-node GPU cluster: a control plane (login, identity, the Slurm controller, and Docker) on a single VMware host, with a NAS and the GPU nodes alongside.
The platform:
- Identity: FreeIPA, one login for the whole group
- Scheduling: Slurm with a fair queue
- Network: a 10 GbE MikroTik LAN linking the NAS and control plane
- Storage: TrueNAS shared storage