~/homelab

homelab.

three nixos boxes on tailscale. runs everything i refuse to cloud — photos, media, matrix, push notifications. host telemetry pushed every 60s by the homelab-agent module on each box; service status scraped from uptime-kuma on nova.

hosts 3/3 reportingservices 9 · 4 down

01 //topology.

ascii · the whole rack in 20 lines

     [ internet ]
          │
     ┌────┴────┐
     │ router  │   home gateway · 192.168.0.0/24
     └────┬────┘
          │
     ┌────┴───────────────┬──────────────┐
     │                    │              │
 ┌───┴────┐           ┌───┴────┐    ┌────┴───┐
 │  nova  │           │gilbert │    │  void  │
 │  .10   │           │  .11   │    │  .12   │
 │ caddy  │           │ ripper │    │  zfs   │
 │ matrix │           │  mc    │    │  nas   │
 │ media  │           │  nfs   │    │ samba  │
 └───┬────┘           └────────┘    └────────┘
     │
     ▼
 [ tailscale ⇄ cloudflare dns ⇄ *.flaked.org ]

02 //hosts.

3 machines · all nixos · all on tailscale

novamedia server / reverse proxy / matrix

cpu: IntelN100 · 4c
ram: 15.4 GiB
storage: root: 47.7 GiB / 467.9 GiB
os: nixos 26.05
load: 0.20
uptime: 36d 2h
last seen: 44s ago

gilbertmedia ripping / minecraft / nfs

cpu: Inteli7-6700 CPU @ 3.40GHz · 8c
ram: 15.3 GiB
storage: root: 56.6 GiB / 232.7 GiB
os: nixos 26.05
load: 0.15
uptime: 90d 4h
last seen: 52s ago

voidnas · zfs raid

cpu: Inteli5-4460 CPU @ 3.20GHz · 4c
ram: 15.5 GiB
storage: storage: 2.72 TiB ONLINE
os: nixos 26.05
load: 0.73
uptime: 89d 6h
last seen: 10s ago

03 //services.

9 containers across 3 hosts

jellyfinmedia@nova

self-hosted media library; streams whatever gilbert ripped to whichever tv is on

statusup

uptime100.00%

latency70ms

immichmedia@nova

photo library, replaces icloud for anything i actually care about

statusup

uptime100.00%

latency0ms

matrix-synapseinfra@nova

matrix homeserver; federates out, postgres-backed

statusup

uptime100.00%

latency95ms

piholesecurity@nova

network-wide dns blocklists for the whole house

statusup

uptime100.00%

latency97ms

uptime-kumainfra@nova

monitors every other service. this page reads from its api.

statusup

uptime100.00%

latency65ms

rommmedia@nova

rom library + emulator frontend; postgres-backed

statusup

uptime100.00%

latency0ms

rustfsstorage@nova

s3-compatible object store; serves as a backup target

statusdown

uptime0.00%

latency0ms

gotifyinfra@nova

push notifications for nixos upgrades + alerts

statusup

uptime100.00%

latency0ms

igotifyinfra@nova

second gotify instance — split channel for noisier alerts

statusdown

uptime0.00%

latency0ms

04 //storage.

raid6 · nightly restic → b2

capacity// zfs pools across all hosts

1.3TiB used

total 2.7 TiBfree 1.4 TiB47% used

backups// rustic-backup module

tool: rustic (restic-compatible)
schedule: nightly via systemd timer
encryption: per-host repo password (sops)
last snapshot: tbd — agent will report

network// tailscale + caddy

subnet: 192.168.0.0/24
tailscale: nova as exit + subnet router
public dns: *.flaked.org via cloudflare
dns blocking: tbd — pihole stats