Last updated: July 29, 2026Latest Frontend Version:

Slurm Administration

tiCrypt uses a dual-Slurm architecture. Global Slurm runs at the platform level and schedules VM provisioning across physical hosts. It never sees job content, job scripts, or data. Local Slurm runs inside each encrypted enclave, and is the instance researchers interact with when they run sbatch, srun, or squeue. From the researcher's perspective, the commands and output are identical to a traditional cluster.

This section covers the administrative side: configuring the Slurm cluster, managing nodes and worker VMs, monitoring jobs, and managing user and VM accounts.

info

For background on how this architecture works and why tiCrypt splits Slurm into two layers, see Shared Computing to tiCrypt. For the security design, see Secure HPC with Slurm.

Page	What it covers
Slurm Configuration	Cluster setup, partitions, and node configuration
Nodes	Viewing and managing Slurm compute nodes
Worker VMs	Ephemeral VMs created by Global Slurm for batch jobs
Active Jobs	Monitoring currently running and queued jobs
Past Jobs	Historical job records and resource consumption
User Accounts	Slurm user account management
VM Accounts	VM-level Slurm account configuration
Slurm Diagnostics	Cluster health and diagnostic information