core/go
8
0
Fork 0
generated from lthn/LEM
Code Issues 5 Pull requests 1 Projects Releases Packages Wiki Activity Actions

docs: update .lan service URLs to *.lthn.lan subdomain convention

Browse source 
All homelab services now use the *.lthn.lan naming convention
(ollama.lthn.lan, qdrant.lthn.lan, eaas.lthn.lan) per updated
/etc/hosts configuration.

Co-Authored-By: Virgil <virgil@lethean.io>
This commit is contained in:
Snider 2026-03-03 13:19:57 +00:00
parent bb9844c638
commit 7fe48a6268
2 changed files with 18 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
docs/plans/2026-03-03-lthn-lan-handover.md
View file

@ -6,7 +6,7 @@
## Goal
Stand up the Host UK Laravel app on the Linux homelab as `lthn.lan` — a private dev/ops hub away from production. This joins the existing `.lan` service mesh (ollama.lan, qdrant.lan, eaas.lan).
Stand up the Host UK Laravel app on the Linux homelab as `lthn.lan` — a private dev/ops hub away from production. This joins the existing `.lan` service mesh (ollama.lthn.lan, qdrant.lthn.lan, eaas.lthn.lan).
## What lthn.lan Is
@ -22,9 +22,9 @@ Mac (snider) ──hosts file──▶ lthn.lan (10.69.69.165)
└── Redis/Dragonfly (port 6379)
Already running on 10.69.69.165:
ollama.lan → Ollama (embeddings, LEM inference)
qdrant.lan → Qdrant (vector search)
eaas.lan → EaaS scoring API v0.2.0
ollama.lthn.lan → Ollama (embeddings, LEM inference)
qdrant.lthn.lan → Qdrant (vector search)
eaas.lthn.lan → EaaS scoring API v0.2.0
```
## Prerequisites
@ -147,13 +147,13 @@ BROADCAST_CONNECTION=log
OCTANE_SERVER=frankenphp
# OpenBrain — connects to existing .lan services
BRAIN_OLLAMA_URL=https://ollama.lan
BRAIN_QDRANT_URL=https://qdrant.lan
BRAIN_OLLAMA_URL=https://ollama.lthn.lan
BRAIN_QDRANT_URL=https://qdrant.lthn.lan
BRAIN_COLLECTION=openbrain
BRAIN_EMBEDDING_MODEL=embeddinggemma
# EaaS scorer
EAAS_URL=https://eaas.lan
EAAS_URL=https://eaas.lthn.lan
```
Then generate the app key:
@ -175,7 +175,7 @@ labels:
traefik.docker.network: proxy
```
Note: For `.lan` domains, Traefik uses self-signed certs (no Let's Encrypt — not a real TLD). The same pattern as ollama.lan/qdrant.lan/eaas.lan.
Note: For `.lan` domains, Traefik uses self-signed certs (no Let's Encrypt — not a real TLD). The same pattern as ollama.lthn.lan/qdrant.lthn.lan/eaas.lthn.lan.
## Step 5: Build and Start
@ -225,11 +225,11 @@ Already done by snider:
## Embedding Model on GPU
The `embeddinggemma` model on ollama.lan appears to be running on CPU. It's only ~256MB — should fit easily alongside whatever else is on the RX 7800 XT. Check with:
The `embeddinggemma` model on ollama.lthn.lan appears to be running on CPU. It's only ~256MB — should fit easily alongside whatever else is on the RX 7800 XT. Check with:
```bash
# On the Linux machine
curl -sk https://ollama.lan/api/ps
curl -sk https://ollama.lthn.lan/api/ps
```
If it shows CPU, try pulling it fresh or restarting Ollama — it should auto-detect the GPU.

16
docs/plans/2026-03-03-openbrain-usage.md
View file

@ -25,8 +25,8 @@ Agent ──recall()────▶ BrainService
| Service | URL | What |
|---------|-----|------|
| Ollama | `https://ollama.lan` | Embedding model (`embeddinggemma`, 768 dimensions) |
| Qdrant | `https://qdrant.lan` | Vector storage + cosine similarity search |
| Ollama | `https://ollama.lthn.lan` | Embedding model (`embeddinggemma`, 768 dimensions) |
| Qdrant | `https://qdrant.lthn.lan` | Vector storage + cosine similarity search |
| MariaDB | `lthn-lan-db:3306` | `brain_memories` table (workspace-scoped) |
| Laravel | `https://lthn.lan` | BrainService, artisan commands, MCP tools |
@ -80,8 +80,8 @@ If the Laravel app isn't available, use the Go brain-seed tool:
```bash
cd ~/Code/go-ai
go run cmd/brain-seed/main.go \
--ollama=https://ollama.lan \
--qdrant=https://qdrant.lan \
--ollama=https://ollama.lthn.lan \
--qdrant=https://qdrant.lthn.lan \
--collection=openbrain \
--model=embeddinggemma
```
@ -134,14 +134,14 @@ For debugging or bulk operations:
```bash
# Collection stats
curl -sk https://qdrant.lan/collections/openbrain | python3 -m json.tool
curl -sk https://qdrant.lthn.lan/collections/openbrain | python3 -m json.tool
# Raw vector search (embed query first via Ollama)
VECTOR=$(curl -sk https://ollama.lan/api/embeddings \
VECTOR=$(curl -sk https://ollama.lthn.lan/api/embeddings \
-d '{"model":"embeddinggemma","prompt":"Traefik setup"}' \
| python3 -c "import sys,json; print(json.dumps(json.load(sys.stdin)['embedding']))")
curl -sk https://qdrant.lan/collections/openbrain/points/search \
curl -sk https://qdrant.lthn.lan/collections/openbrain/points/search \
-H 'Content-Type: application/json' \
-d "{\"vector\":$VECTOR,\"limit\":5,\"with_payload\":true}" \
| python3 -m json.tool
@ -215,7 +215,7 @@ php artisan brain:ingest --workspace=1 --fresh --source=memory
### Check Collection Health
```bash
curl -sk https://qdrant.lan/collections/openbrain | \
curl -sk https://qdrant.lthn.lan/collections/openbrain | \
python3 -c "import sys,json; r=json.load(sys.stdin)['result']; print(f'Points: {r[\"points_count\"]}, Status: {r[\"status\"]}')"
```