GNAI-008/M4: Production-Inventar mit Referenz- und Rollback-Profil

inventories/production: eigener Workload-Account gnai-production (980),
Quadlet-Verzeichnis, Unit-Praefix gnai-production-llama und Podman-Netz.
Als Profile nur der freigegebene Referenz Swift 1.5 Q6_K (Betreiber-
beschluss 05.11.2026, GNAI-009 10.4) sowie Unsloth UD-Q6_K als Rollback.
Preview: production changed=2, testing-Regressionslauf changed=0.

Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>
This commit is contained in:
2026-11-06 06:34:05 +01:00
parent 95cca54aef
commit 4653fb573a
2 changed files with 68 additions and 0 deletions
@@ -0,0 +1,61 @@
---
# Environment-specific values for the Production stack (GNAI-008 section
# 7.4, section 7.2 isolation table; GNAI-001 milestone 4). Same role logic
# as Testing with its own workload account, quadlet directory, unit prefix
# and network. Production deploys only the approved reference profile
# (Swift 1.5 Q6_K, operator decision 2026-11-05, GNAI-009 10.4) plus the
# rollback profile (Unsloth UD-Q6_K); profile parameters are kept in sync
# with the role defaults until the promotion chain (GNAI-008 section 12)
# automates the release.
inference_engine_user: gnai-production
inference_engine_uid: 980
inference_engine_user_home: /var/lib/ghostnet-ai/production
inference_engine_unit_prefix: gnai-production-llama
inference_engine_network_name: gnai-production-llama
inference_engine_network_alias: llama-cpp
# Approved reference and rollback profile only (see the comment above).
inference_engine_profiles:
swift-1.5-qwen3.8-27b-q6k:
unit_suffix: swift15
description: Swift 1.5 reasoning-efficient finetune of Qwen3.8-27B (Q6_K)
model_directory: swift-1.5-qwen3.8-27b
model_file: Swift-1.5-Qwen3.8-27B-Q6_K.gguf
alias: swift-1.5-qwen3.8-27b-q6k
ctx_size: 131072
n_predict: 32768
parallel: 1
reasoning_effort: xhigh
temp: 1.0
top_p: 0.95
top_k: 20
min_p: 0.0
presence_penalty: 0.0
repeat_penalty: 1.0
mtp:
enabled: false
spec_draft_n_max: 2
unsloth-qwen3.8-27b-ud-q6k:
unit_suffix: unsloth
description: Unsloth UD quant of Qwen3.8-27B (Q6_K)
model_directory: unsloth-qwen3.8-27b
model_file: Qwen3.8-27B-UD-Q6_K.gguf
alias: unsloth-qwen3.8-27b-ud-q6k
ctx_size: 131072
n_predict: 32768
parallel: 1
reasoning_effort: xhigh
temp: 1.0
top_p: 0.95
top_k: 20
min_p: 0.0
presence_penalty: 0.0
repeat_penalty: 1.0
mtp:
enabled: false
spec_draft_n_max: 2
inference_engine_active_profile: swift-1.5-qwen3.8-27b-q6k
# 128k target context, same as Testing (GNAI-010 section 10.9).
inference_engine_ctx_size_override: null
@@ -0,0 +1,7 @@
all:
children:
llm_servers:
hosts:
ghostnet-ki:
ansible_host: 10.0.0.90
ansible_user: ansible-deploy