Neues Kontrollprojekt llm-server/llm/ansible (GNAI-008 7.4, GNAI-010): - Inventar testing (ansible-deploy), Playbook ai.yml - Rolle model_store: gepinnte Modellartefakte (Swift/Unsloth Q6_K, Revision und SHA-256 nach GNAI-010 3.2/3.3), atomare Downloads mit Prüfsummen, 70-GiB-Freiraumreserve, SOURCE.yml-Herkunftsnachweis je Verzeichnis - Rolle inference_engine: gepinntes ROCm-Image per Digest, internes Quadlet-Netz (Internal=true), zwei gegenseitig ausgeschlossene llama-server-Quadlets (Conflicts/Before nach GNAI-010 5.3), 64k-Bring-up- Override, Generator-Dry-Run, exklusiver Profilwechsel, Health- und Alias-Verifikation; kein Host-Port, kein default.target-Enable - Syntax-Check und Template-Render-Verifikation bestanden; Live-Check-Modus steht aus (KI-Server aktuell ausgeschaltet) Generated by Mistral Vibe. Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
LLM server ghostnet-ki
This host root contains the executable infrastructure for the physical server at
10.0.0.90, split into one main directory per task area:
| Directory | Purpose |
|---|---|
host/ |
shared server baseline: Ansible control project, host roles and inventories |
llm/ |
LLM serving stack (created with the internal AI vertical slice) |
encoding/ |
CPU-based video encoding workload and its Ansible control project |
monitoring/ |
host-wide monitoring: metrics collector and dashboard |
Reusable roles that are also consumed by other host projects live under
../shared/ansible/.
The versioned architecture and operating documentation remains under
../docs/platform/llm-server/ and
mirrors the same task areas (host/, ai/, encoding/).