HPC Admin TUI

A terminal-based administration tool for managing GPU HPC clusters running Slurm, Open OnDemand, and Active Directory/SSSD on Rocky Linux 9.

Built with Textual.

Features

• Slurm Admin — manage users, accounts, QoS policies, and partitions via sacctmgr; bulk-import AD groups; live AD group membership view
• Config Sync — cluster-wide config file sync matrix (slurm.conf, sssd.conf, hosts, limits, sysctl, environment); per-file diff and targeted push; proxy config audit
• Scratch Audit — per-node /scratch scanner with age/size highlighting, active-job protection, and bulk delete
• ZFS Quotas — view and set per-user ZFS quotas on TrueNAS NAS hosts
• NetFreeze Monitor — SSH-based network freeze detection across cluster nodes (ARP failure, bond failover, link flap)

Requirements

• Python 3.11+
• pip install textual asyncssh
• Slurm CLI tools (sacctmgr, sinfo, squeue) in PATH
• ldapsearch for AD integration
• SSH key-based access to all cluster nodes

Usage

python3 hpc_admin_tui.py           # live mode
python3 hpc_admin_tui.py --demo    # demo mode (no real commands run)

On first launch you will be prompted for AD credentials. These are held in memory only and never written to disk.

Configuration

Edit the constants at the top of the file before use:

• SSH_USER — admin username for SSH/SCP
• LDAP_URI / LDAP_BASE — your AD domain and search base
• PROMETHEUS — Prometheus host for filesystem metrics
• ZFS_QUOTA_DATASETS — TrueNAS hostnames and dataset paths
• NF_DEFAULT_CONFIG — node list for NetFreeze monitoring