Workstation Ops

Diagnose, monitor, and maintain this workstation.

This Machine

| Component | Spec | |-----------|------| | Motherboard | MSI TRX40 PRO 10G | | CPU | AMD Threadripper 3960X (24c/48t) | | RAM | 256 GB DDR4 | | GPU | NVIDIA RTX A5000 (24GB) | | Boot | WD Black SN850X 4TB → / | | Storage | Seagate IronWolf 12TB → /mnt/storage12tb |

Commands

| Command | What it does | |---------|--------------| | ./run.sh | Quick resource check | | ./run.sh memory | What's using RAM | | ./run.sh gpu | GPU status & VRAM | | ./run.sh health | Drive health & maintenance status | | ./run.sh net | Network diagnostics (link/IP/DNS/sockets) | | ./run.sh temps | Temperature monitoring (CPU/GPU/NVMe) | | ./run.sh containers | Docker container health | | ./run.sh slim | Find storage savings (media, caches, models) | | ./run.sh duplicates | Find exact duplicate files | | ./run.sh crashes | Check for OOM/crashes | | ./run.sh specs | Hardware docs & upgrade procedures |

Agent Quick Reference

"Is the system healthy?"

./run.sh health

"What's using all the memory?"

./run.sh memory

"Will my model fit in GPU?"

./run.sh gpu --need 8000

"Something crashed"

./run.sh crashes --oom

"Where can I recover storage?"

./run.sh slim                      # Overview of all opportunities
./run.sh slim --media              # Lower quality media versions
./run.sh slim --caches             # Cache cleanup (pip, docker, npm)
./run.sh slim --models             # ML model caches

"Is the network OK?"

./run.sh net                       # LAN, DNS, sockets, gateway
./run.sh net --json                # Agent-parseable JSON
./run.sh net --no-external         # Skip external ICMP checks
OUTPUT=json ./run.sh net           # Alternative JSON output

"Are temps safe?"

./run.sh temps                     # Markdown summary (exit code: 0=OK, 1=WARN, 2=CRIT)
./run.sh temps --json              # JSON with exit codes
./run.sh temps --warn 70 --crit 80 # Custom thresholds

"How are my containers?"

./run.sh containers                # Docker status, restart counts
./run.sh containers --json         # Agent-parseable JSON

"Find duplicate movies"

./run.sh duplicates --dry-run              # Quick check by size
./run.sh duplicates                         # Full scan with checksums
./run.sh duplicates --type mkv --min-size 500  # Only large .mkv files

"What can I clean up?"

./run.sh health --cleanup

"How do I add a hard drive?"

./run.sh specs --procedures

Maintenance Recommendations

The skill will identify these issues automatically:

| Issue | How to check | How to fix | |-------|--------------|------------| | High memory | ./run.sh memory | Close IDEs, kill idle Claude Code | | Disk filling up | ./run.sh slim | Clear caches, remove duplicates | | Lower quality media | ./run.sh slim --media | Delete lower quality versions | | Large caches | ./run.sh slim --caches | pip cache purge, docker prune | | Duplicate files | ./run.sh duplicates | Review and remove exact duplicates | | Drive failing | sudo ./run.sh health --drives | Replace drive | | Pending updates | ./run.sh health | sudo apt upgrade |

Quick Cleanup

# See all storage savings opportunities
./run.sh slim

# Caches (can recover 30+ GB)
pip cache purge
npm cache clean --force
docker system prune -a
huggingface-cli delete-cache  # ML models

# System
sudo apt autoremove -y && sudo apt clean
sudo journalctl --vacuum-size=500M

# Find lower quality media versions
./run.sh slim --media

Hardware Reference

See WORKSTATION.md for:

Full hardware specs with mermaid diagrams
Drive layout and available slots
Step-by-step upgrade procedures (NVMe, SATA, RAM, GPU)