System
The machines this gets made on.
One laptop does the work — the writing, the mail server work, the Go and the Node — and a rack in the house is where that work gets to fail somewhere it does not matter. All are LINUX boxes configured by somebody who has to fix them, which is the only kind of recommendation worth anything.
~/jcoffey$ fastfetch
# machine
Laptop: System76 Pangolin Pro (panp16)
CPU: AMD Ryzen AI 7 350 — 8 cores / 16 threads, boost 5.09 GHz
GPU 1: AMD Radeon RX 9070 XT — discrete, amdgpu
GPU 2: AMD Radeon 860M Graphics — integrated, amdgpu
Graphics: Mesa 26.2.1-arch3.1
Memory: 92 GiB
Displays: 2 × Sceptre O34 — 3440×1440 @ 180 Hz, 34in
# system
OS: CachyOS
Kernel: Linux 7.2.2-1-cachyos
Packages: 1687 (pacman)
Desktop: KDE Plasma 6.7.4 on KWin (Wayland)
Shell: fish 4.8.1
Terminal: Konsole 26.08.0
# toolchain
Git: 2.55.0
Node: 26.7.0 / npm 12.0.2
Python: 3.14.7
Go: 1.27.0
Rust: 1.98.0
C / C++: gcc 16.2.1, clang 22.1.8
Docker: 29.7.2
Every day
The commands my hands type without asking me first.
ghrgfdfzfbatezajqclaude
The homelab
None of the writing happens here — that is the laptop above. This is the rack: a fully populated blade chassis, a storage node and a NAS tower, on Proxmox. It is where a thing gets to fail in a way that costs me an evening instead of somebody else a morning.
~/jcoffey$ pvecm status
# cluster
Hypervisor: Proxmox VE
Backplane: 10G Ethernet
# network — top of rack
Router: OPNsense
Switch: MikroTik
Patching: breakout panel
# storage — nas tower (dell poweredge t110 ii)
OS: FreeNAS
CPU: 1× 4-core
Memory: 32 GiB
Storage: 4× 16TB
Capacity: 64 TB raw / ~50 TB combined
# compute — tower (pve2, dell precision t5500)
Node type: Precision tower
CPU: 2× 12-core (48 threads)
Memory: 96 GiB
Storage: 16TB SATA
Role: Proxmox cluster member
# compute — blades (dell m1000e, fully populated)
Blade count: 16
Node type: PowerEdge blade
CPU: 2× 12-core per blade (48 threads)
Memory: 64 GiB per blade
Storage: 2× 1TB SAS per blade
Subtotal: 384 cores / 768 threads / 1,024 GiB RAM / 32 TB SAS (raw)
# compute — storage node (google-branded r710)
CPU: 2× 12-core (48 threads)
Memory: 368 GiB
Storage: 24× 1TB SAS + 512GB SATA cache
Subtotal: 24 cores / 48 threads / 368 GiB RAM / 24 TB SAS (raw) + 512 GB cache
# cluster totals (blades + r710, excl. tower/nas)
Cores: 408
Threads: 816
Memory: 1,392 GiB
Raw SAS: 56 TB
# power — foot of rack
UPS: Tripp Lite
The long versions
- Why I’m Moving from My Home-Built 5800X3D Desktop to a System76 Pangolin Pro why this laptop
- A Debian Veteran Tries CachyOS why this distro