The flash-attention + q8_0 KV cache settings were removed after they (and the gfx1103 ROCm path generally) crash on large prompts. Document that the 780M iGPU is direct/interactive-chat only, not an agent backend (Hermes ~15.5k-token prompt crashed every call; reverted to cloud model). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
12 KiB
Other Services
Last Updated: 2026-06-29
Non-critical services that enhance functionality but don't affect core network operation.
DevOps
Gitea (Git Server)
| Component | IP | URL |
|---|---|---|
| gitea | 172.18.0.31 | https://git.xtrm-lab.org |
Ports: 2222→22 (SSH), 3005→3000 (HTTP) Auth: Authentik OAuth2 Database: postgresql17 (gitea)
Woodpecker CI
| Component | IP | URL |
|---|---|---|
| woodpecker-server | 172.18.0.32 | https://ci.xtrm-lab.org |
| woodpecker-agent | 172.18.0.33 | - |
Auth: Gitea OAuth2 Purpose: CI/CD pipelines for infrastructure repo
Network Management
NetDisco (Network Discovery)
| Component | IP | URL |
|---|---|---|
| netdisco-web | 172.18.0.41 | https://netdisco.xtrm-lab.org |
| netdisco-backend | 172.18.0.42 | - |
Database: postgresql17 (netdisco_db) Purpose: SNMP-based device discovery, MAC/ARP tracking
Unimus
| Network | URL |
|---|---|
| host | https://unimus.xtrm-lab.org |
Purpose: Network device configuration backup (MikroTik)
Monitoring
Uptime Kuma
| Component | IP | URL |
|---|---|---|
| UptimeKuma | 172.18.0.20 | https://uptime.xtrm-lab.org |
| Uptime-Kuma-API | 172.18.0.18 | - |
| AutoKuma | 172.18.0.19 | - |
Monitors: 24 services configured
Speedtest Tracker
| Component | IP | URL |
|---|---|---|
| speedtest-tracker | 172.18.0.21 | https://speed.xtrm-lab.org |
NetAlertX
| Network | URL |
|---|---|
| host | https://netalert.xtrm-lab.org |
Purpose: Network device discovery and alerting
Gaming
Minecraft Server
| Component | IP | Port | Address |
|---|---|---|---|
| minecraft | 172.18.0.80 | 25565 | minecraft.xtrm-lab.org |
Image: itzg/minecraft-server (Java Edition)
Version: Latest (1.21.11)
Mode: Survival, Normal difficulty, PVP enabled
Max Players: 10
RAM: 2 GB
Online Mode: Yes (requires paid account)
Data: /mnt/user/appdata/minecraft/data
NAT: WAN:25565 → 192.168.10.20:25565 + hairpin NAT
Dockge Stack: minecraft
Media
Plex
| Network | Port | URL |
|---|---|---|
| host | 32400 | https://plex.xtrm-lab.org |
Libraries: Movies, TV Shows, Music
Plex Debrid
| Network | Purpose |
|---|---|
| host | Debrid integration for Plex |
Purpose: Integrates Real-Debrid/AllDebrid with Plex for streaming
Zurg + Rclone
| Component | Port | Purpose |
|---|---|---|
| zurg | 9999 | Real-Debrid WebDAV mount |
| zurg-rclone | - | Rclone FUSE mount |
Purpose: Mounts Real-Debrid library as local filesystem for Plex
Overseerr
| Component | Port | URL |
|---|---|---|
| overseerr | 5055 | https://overseerr.xtrm-lab.org |
Purpose: Media request management for Plex
Radarr
| Component | IP | URL |
|---|---|---|
| radarr | 172.18.0.43 | https://radarr.xtrm-lab.org |
Purpose: Movie collection manager (PVR)
Sonarr
| Component | IP | URL |
|---|---|---|
| sonarr | 172.18.0.44 | https://sonarr.xtrm-lab.org |
Purpose: TV series collection manager (PVR)
Prowlarr
| Component | IP | URL |
|---|---|---|
| prowlarr | 172.18.0.45 | https://prowlarr.xtrm-lab.org |
Purpose: Indexer manager for Radarr/Sonarr
Libation
| Component | IP |
|---|---|
| Libation | 172.18.0.23 |
Purpose: Audiobook management
Transmission
| Component | IP | Ports |
|---|---|---|
| transmission | 172.18.0.26 | 9091, 51413 |
Purpose: Torrent client
Roms (SMB Share)
| Property | Value |
|---|---|
| Share Path | /mnt/user/roms |
| Protocol | SMB (guest access, read-only) |
| Size | 2.3 TB (49 systems) |
Consumers:
| Device | Mount Point | Method |
|---|---|---|
| Nobara | /mnt/roms | fstab (CIFS, guest, systemd.automount) |
| Recalbox | /recalbox/share/roms_network | custom.sh boot script (CIFS) |
Recalbox: Network roms are bind-mounted over local rom directories at boot via /recalbox/share/system/custom.sh. Local roms were deleted from SD card to save space.
Productivity
Nextcloud
| Component | IP | URL |
|---|---|---|
| Nextcloud | 172.18.0.24 | https://cloud.xtrm-lab.org |
Purpose: File sync, calendar, contacts
Actual Budget
| Component | IP | URL |
|---|---|---|
| actual-budget | 172.18.0.16 | https://actual.xtrm-lab.org |
Purpose: Personal finance tracking
n8n
| Component | IP | URL |
|---|---|---|
| n8n | 172.18.0.17 | https://n8n.xtrm-lab.org |
Purpose: Workflow automation
Karakeep
| Component | IP | URL |
|---|---|---|
| karakeep | 172.18.0.25 | https://karakeep.xtrm-lab.org |
Purpose: Bookmark manager
Homarr (Dashboard)
| Component | IP | URL |
|---|---|---|
| homarr | 172.18.0.4 | https://xtrm-lab.org |
Purpose: Home dashboard with Docker integration
Storage & Backup
TimeMachine
| Network | IP |
|---|---|
| br0 macvlan | 192.168.10.12 |
Purpose: macOS Time Machine backup target
UrBackup
| Network | URL |
|---|---|
| host | https://urbackup.xtrm-lab.org |
Purpose: File and image backup for PCs
RustFS
| Network |
|---|
| bridge |
Purpose: Rust-based filesystem utilities
Remote Access
Kasm Workspaces (Browser Isolation)
| Component | IP | URL |
|---|---|---|
| kasm (DinD host) | 172.18.0.66 | https://kasm.xtrm-lab.org |
| kasm_proxy | 172.19.0.8 (internal) | - |
| kasm_api | 172.19.0.7 (internal) | - |
| kasm_manager | 172.19.0.6 (internal) | - |
| kasm_agent | 172.19.0.2 (internal) | - |
| kasm_db | 172.19.0.3 (internal) | - |
| kasm_guac | 172.19.0.9 (internal) | - |
Version: 1.18.1
Architecture: Docker-in-Docker (runs its own Docker daemon with internal containers)
Network: dockerproxy (outer), kasm_default_network 172.19.0.0/16 (inner)
Ports: 3000 (web UI), 6333 (proxy/sessions)
Data: /mnt/user/appdata/kasm/
Traefik Integration:
- Uses
kasm-headersmiddleware (noframeDeny— Kasm requires iframes) - Service backend:
https://172.18.0.66:6333withinsecure-skip-verify - Zone config:
proxy_port=443,proxy_hostname=$request_host$(routes sessions through Traefik)
RustDesk
| Component | Ports |
|---|---|
| rustdesk-hbbs | 21115-21116 (TCP), 21116 (UDP) |
| rustdesk-hbbr | 21117-21119 |
Purpose: Self-hosted remote desktop
Tailscale
| Host | IP |
|---|---|
| HAP1 | 172.17.0.3 |
Purpose: Mesh VPN for remote access
Open WebUI
| Component | Port |
|---|---|
| open-webui | 3080 |
Purpose: Web UI for Ollama (ChatGPT-like interface)
Smart Home
Home Assistant
| Network | URL |
|---|---|
| host | https://ha.xtrm-lab.org |
Purpose: Home automation hub Trusted Proxies: 172.18.0.0/16, 192.168.10.0/24
Home Assistant Time Machine
| Component | Port |
|---|---|
| Home-Assistant-Time-Machine | 54000 |
Purpose: HA backup/snapshot management
Tuya Gateway
| Network |
|---|
| host |
Purpose: Local Tuya device integration
Notifications
ntfy
| Component | Port | Purpose |
|---|---|---|
| ntfy | 8380 | Push notification server |
Purpose: Self-hosted push notifications (UnifiedPush compatible)
Sync & Knowledge
Obsidian LiveSync
| Component | Port |
|---|---|
| obsidian-livesync | 5984 |
Purpose: CouchDB-based real-time sync for Obsidian vaults
Supabase (OpenBrain)
| Component | Purpose |
|---|---|
| supabase-db | PostgreSQL database |
| supabase-auth | Authentication |
| supabase-rest | PostgREST API |
| supabase-storage | File storage |
| supabase-studio | Web dashboard |
| supabase-meta | Metadata service |
| supabase-analytics | Analytics/logging |
| supabase-imgproxy | Image processing |
| supabase-edge-functions | Edge functions runtime |
| supabase-pooler | Connection pooling |
| supabase-vector | Vector/embeddings |
| supabase-kong | API gateway (stopped) |
| realtime-dev.supabase-realtime | Realtime subscriptions |
Purpose: Self-hosted Supabase stack with PGVector for AI persistent memory (OpenBrain)
Database Tools
pgAdmin 4
| Component | Port |
|---|---|
| pgAdmin4 | 5050 |
Purpose: PostgreSQL web administration UI
Development
OpenVSCode Server (Web IDE)
| Network | Port | URL |
|---|---|---|
| host | 3100 | https://code.xtrm-lab.org |
Purpose: Web-based VS Code IDE with full host access
Auth: Authentik forward auth (SSO)
Binary: /mnt/user/appdata/openvscode/current/ (host-native, not a container)
Config: /mnt/user/appdata/openvscode/config/
Workspace: /mnt/user/projects/ (12 personal repos + 18 AMPECO projects)
See: 12-DEVELOPMENT-ENVIRONMENT.md
Claude Code
| Network | Version |
|---|---|
| host | 2.1.71 |
Purpose: AI coding assistant CLI
Binary: /mnt/user/appdata/claude-code/.npm-global/bin/claude
Config: /mnt/user/appdata/claude-code/.claude/
Cooperator CLI
| Network | Version |
|---|---|
| host | 3.36.1 |
Purpose: AI-driven development workflow automation (AMPECO)
Binary: /usr/local/lib/node_modules/@ampeco/cooperator/
Container Management
Dockge
| Network | Port |
|---|---|
| bridge | 5001 |
Purpose: Docker Compose stack management UI
Access: http://192.168.10.20:5001
Data: /mnt/user/appdata/dockge/data
Stacks: /mnt/user/appdata/dockge/stacks
Dockhand
| Network |
|---|
| dockerproxy |
Purpose: Docker container health monitoring
Docker Socket Proxy
| Component | IP |
|---|---|
| dockersocket | 172.18.0.2 |
Purpose: Secure Docker socket proxy for Traefik and other services
Agents
xtrm-agent
| Component | Port |
|---|---|
| xtrm-agent | 8088 |
Purpose: Claude Code agent for server management
SeekAndWatch
| Component |
|---|
| SeekAndWatch |
Purpose: Media monitoring/automation
AI / LLM
Ollama (Local LLM Inference)
| Network | Port |
|---|---|
| bridge | 11434 |
Purpose: Local LLM inference, GPU-accelerated on the integrated AMD Radeon 780M (gfx1103) via ROCm 7.2.
Image: ollama/ollama:rocm (Dockge stack — replaced old dockerman template my-ollama.xml)
API: http://192.168.10.20:11434
Data: /mnt/user/appdata/ollama
Stack: /mnt/user/appdata/dockge/stacks/ollama/compose.yaml
iGPU enablement (gfx1103 not on AMD's official ROCm allowlist):
| Setting | Value |
|---|---|
HSA_OVERRIDE_GFX_VERSION |
11.0.0 (report 780M as gfx1100) |
OLLAMA_IGPU_ENABLE |
1 (Ollama 0.30+ drops iGPUs otherwise) |
| devices | /dev/kfd, /dev/dri |
| group_add | 18 (video) |
Verified: Ornith 1.0 9B GGUF runs 100% on GPU (33/33 layers), ~11 tok/s. iGPU is RAM-bandwidth bound (shared GTT, ~23 GiB addressable). 9B fits; 35B RAM-tight (no swap); 397B not feasible.
⚠️ Limitation: large prompts (~10k+ tokens) crash the iGPU (ROCm unspecified launch failure — gfx1103 rocBLAS). Tried flash-attn off, gfx1100/gfx1102, num_batch=128 — all crash, so OLLAMA_FLASH_ATTENTION/OLLAMA_KV_CACHE_TYPE=q8_0 were removed. Good for direct/interactive chat only, not an agent backend with big system prompts (tried as Hermes' model with a ~15.5k-token prompt — crashed every call, reverted to cloud).
Stopped/Disabled Services
| Service | Reason | Status |
|---|---|---|
| charge-* (12 containers) | Dev/test stack, stopped 6 weeks ago | Exited |
| minecraft | Not started | Created |
| supabase-kong | API gateway, stopped 2 days ago | Exited |
| unmarr | Stopped 5 days ago | Exited |
| llama-baseline-test | LLM benchmark, completed | Exited |
| llama-spec-test | LLM benchmark, completed | Exited |
| NetBox + Diode (8 containers) | Removed from active use | Removed |
| Portainer | Replaced by Dockge | Removed |
| Pi-hole | Replaced by AdGuard Home | Removed |
| Slurp'it | Replaced by Diode | Removed |
| binhex-plexpass | Duplicate of Plex | Removed |
| HomeAssistant_inabox | Duplicate of Home-Assistant-Container | Removed |
| nextcloud-aio-mastercontainer | Replaced by Nextcloud container | Removed |