fix: restore files unintentionally reverted by sync commit

- restore jgpc/ layout (undo accidental revert to thinkstation/), homelab-overview.md
  (June 13 version), docs/jarvis doc, owncloud scripts, siklos/beszel compose
- re-remove jellyfin/node-exporter and siklos/monitoring (replaced by Beszel)
- jgpc/frigate/docker-compose.yml: update to newer local version (cpuset pinning, 8G limit)
- root frigate-* files are local working copies -> gitignored
This commit is contained in:
2026-07-08 14:13:20 -05:00
parent 7cb2ebed2d
commit 92c96dc678
15 changed files with 375 additions and 121 deletions
+21 -6
View File
@@ -1,6 +1,6 @@
# Homelab Overview
**Last Updated:** May 11, 2026
**Last Updated:** June 13, 2026
This document provides a high-level summary of the entire homelab infrastructure — networking, physical/virtual hosts, Docker services, and how everything is connected.
@@ -81,7 +81,7 @@ Proxmox runs all the VMs and LXC containers listed below. Backups are handled by
| Name | IP | Role |
|---|---|---|
| thinkstation-p710 | 192.168.88.41 | Workstation / AI / Ollama (bare metal, not a Proxmox guest) |
| jgpc | 192.168.88.41 | Workstation / AI / Ollama (bare metal, not a Proxmox guest) |
| TrueNAS | 192.168.88.24 | NAS — provides NFS storage for Kopia backups; separate physical machine |
---
@@ -107,7 +107,7 @@ Config location on the Caddy VM: `/etc/caddy/`
## Docker Infrastructure
Docker is managed centrally via **Portainer CE** at `http://192.168.88.27:9000` (external: `https://portainer.jgitta.com`). Portainer agents run on jellyfin and thinkstation, allowing all three hosts to be managed from one UI.
Docker is managed centrally via **Portainer CE** at `http://192.168.88.27:9000` (external: `https://portainer.jgitta.com`). Portainer agents run on jellyfin and jgpc, allowing all three hosts to be managed from one UI.
All Docker Compose source files live in the **Gitea** repo `jgitta/homelab-configs` (this repo).
All stack data volumes are stored at `/srv/docker/<stack-name>/` on each host using **absolute paths**.
@@ -120,7 +120,7 @@ All stack data volumes are stored at `/srv/docker/<stack-name>/` on each host us
|---|---|---|---|---|
| Portainer | portainer/portainer-ce:latest | 9000, 9443 | portainer.jgitta.com | Docker management UI |
| Pi-hole | pihole/pihole:latest | 53 (DNS), 8080 | pihole.jgitta.com | DNS ad blocker |
| Open WebUI | ghcr.io/open-webui/open-webui:main | 3000 | ai.jgitta.com | AI chat UI (connects to Ollama on ThinkStation) |
| Open WebUI | ghcr.io/open-webui/open-webui:main | 3000 | ai.jgitta.com | AI chat UI (connects to Ollama on jgpc) |
| LiteLLM | litellm:latest | 4000 | — | Centralized LLM API gateway |
| OnlyOffice | onlyoffice/documentserver | 8880 | office.jgitta.com | Online document editing (used by Nextcloud) |
| Homarr | homarr:latest | 7575 | homarr.jgitta.com | Dashboard |
@@ -164,7 +164,7 @@ Media storage is mounted at `/mnt/media` on the jellyfin host and presented as `
---
### thinkstation-p710 — 192.168.88.41 — AI / Workstation Host
### jgpc — 192.168.88.41 — AI / Workstation Host
| Service | Image | Port | Purpose |
|---|---|---|---|
@@ -174,6 +174,20 @@ Media storage is mounted at `/mnt/media` on the jellyfin host and presented as `
| Stirling PDF | frooodle/s-pdf:latest | 8081 | PDF tools web UI |
| Portainer Agent | portainer/agent:latest | 9001 | Managed by siklos Portainer |
#### Printing (jgpc)
The jgpc is the print server for the HP PageWide Color Flow MFP 586.
| Item | Value |
|---|---|
| Printer | HP PageWide Color Flow MFP 586 |
| IP | 192.168.88.89 (static DHCP reservation, MAC: `ec:8e:b5:c0:d0:48`) |
| CUPS queue | `HP_586_MFP` (default) |
| Device URI | `ipps://192.168.88.89/ipp/print` |
| Driver | IPP Everywhere (driverless) |
`cups-browsed` is **disabled** — it was auto-recreating an unstable mDNS-based queue whenever the printer was online. The printer is permanently saved in `/etc/cups/printers.conf` with a static IP URI.
---
### photos VM — 192.168.88.32 (VM 113) — Photo Services
@@ -190,7 +204,7 @@ Photo storage is on a local 200 GB SSD (`/mnt/immich-photos`) — migrated from
## Monitoring
Prometheus scrapes **Node Exporter** (port 9100) from every host: siklos, proxmox, Nextcloud, jellyfin, PBS, caddy, and ThinkStation. **Grafana** at `grafana.jgitta.com` visualizes all metrics and fires alerts (via Telegram) when:
Prometheus scrapes **Node Exporter** (port 9100) from every host: siklos, proxmox, Nextcloud, jellyfin, PBS, caddy, and jgpc. **Grafana** at `grafana.jgitta.com` visualizes all metrics and fires alerts (via Telegram) when:
- RAM > 90% for 5 minutes
- Swap > 50% for 5 minutes
@@ -228,3 +242,4 @@ Prometheus scrapes **Node Exporter** (port 9100) from every host: siklos, proxmo
| AI (Open WebUI) | https://ai.jgitta.com |
| Guacamole | https://apache.jgitta.com |
| Kopia | https://kopia.jgitta.com |