Compare commits

..

42 Commits

Author SHA1 Message Date
781d98046b refactor: split ai/ into hermes/ and ollama/ directories
- Rename ai/ to hermes/ (Hermes agent service)
- Move ai/ollama/ to ollama/ (top-level, Ollama ROCm/gfx906 service)
- Add ollama/compose.yml for standalone ollama deployment
- Update hermes/compose.yml: remove ollama service (now in ollama/)
- Update Makefile SERVICES list: ai -> hermes, add ollama
- Bump ollama version from v0.13.5 to v0.23.2
2026-05-09 21:44:46 -04:00
ef58155897 feat: add custom ollama image with ROCm 6.1 + gfx906 support
- Add ollama/Dockerfile that builds ollama from source with AMDGPU_TARGETS=gfx906
- Uses ROCm 6.1 (rocm/dev-ubuntu-22.04:6.1.2-complete) for MI50 support
- Builds llama.cpp runner with HIPBLAS for gfx906 architecture
- Updates compose.yml to build from this Dockerfile instead of pulling ollama/ollama:latest
2026-05-09 21:18:37 -04:00
0580603f27 Merge pull request 'fix: add TZ=America/Montreal for correct cron scheduling' (#18) from fix/hermes-timezone-v2 into master
Reviewed-on: #18
2026-05-09 19:54:52 +00:00
93c01fa314 fix: add TZ=America/Montreal for correct cron scheduling 2026-05-09 19:49:54 +00:00
6e540635bf Merge pull request 'feat(hermes): Piper TTS (local US male, no cloud)' (#17) from feat/voice-support-v2 into master
Reviewed-on: #17
2026-05-09 19:39:11 +00:00
Thierry Pouplier
b89be7b8f4 chore: restore system packages lost in base image migration
The migration from debian:stable-slim to nousresearch/hermes-agent:latest
dropped several packages that were previously installed. This restores:

- poppler-utils, imagemagick (PDF/image processing)
- texlive-latex-base, latex-extra, fonts-recommended, xetex, science
- qemu-user-static, binfmt-support (cross-compilation)
- emacs-nox (text editing)

These were added in PRs 3/5, 4/5, 5/5 and earlier commits of the
compose repo. The official image already has git, curl, ffmpeg,
python3, gcc, openssh, ripgrep, tini, docker-cli, etc.
2026-05-09 19:18:16 +00:00
Thierry Pouplier
748b5037b9 fix: update TTS patch for latest hermes-agent tts_tool.py
- Patch now matches the current tts_tool.py (newer version ships in
  nousresearch/hermes-agent:latest with different Edge fallback text)
- Adds dedicated elif provider == 'piper' block before else:
- Replaces else: fallback to use Piper instead of Edge
- Patches ALL copies (venv site-packages + /opt/hermes/tools/)
- Removes Edge TTS entirely as default/provider
2026-05-09 19:03:10 +00:00
Thierry Pouplier
3f80744ebd fix: install piper-tts as root (venv is root-owned in base image)
The nousresearch/hermes-agent:latest image creates its venv
as root. Running 'uv pip install' as USER hermes fails with
Permission denied on the site-packages directory.

Fix: keep USER root while modifying the venv, then switch
back to USER hermes for runtime.
2026-05-09 17:47:30 +00:00
Thierry Pouplier
6f17743667 fix: install into existing venv instead of recreating it
The nousresearch/hermes-agent:latest base image already has a
venv with hermes-agent installed at /opt/hermes/.venv/.
Running 'uv venv' on top of it either fails or wipes the
existing install.

Fix: activate the existing venv first, then pip install into it.
2026-05-09 17:44:55 +00:00
Thierry Pouplier
98216d2872 refactor: use official Hermes Agent image as base, not debian:stable-slim
Starting from debian:stable-slim required re-installing everything
(Hermes source, Node.js, Playwright, etc.) which was redundant
and fragile. The official nousresearch/hermes-agent image already
has all that.

Now the Dockerfile:
- FROM nousresearch/hermes-agent:latest (has tts_tool.py, Playwright, etc.)
- Install Piper + voice model on top
- Patch tts_tool.py at build time (Edge fallback -> Piper)
- Runtime fallback in fix-permissions.sh for volume resilience

Cleaner, smaller Dockerfile, and the build-time patch can find
tts_tool.py because it's in the base image's venv.
2026-05-09 17:39:23 +00:00
Thierry Pouplier
a40e347dfa fix: install hermes-agent from pip so build-time TTS patch works
The Dockerfile starts from debian:stable-slim, not from the official
Hermes image. Without installing hermes-agent from pip, there is no
tools/tts_tool.py in the image at build time, so the patch script
crashes with FileNotFoundError.

Adding hermes-agent to uv pip install gives us tts_tool.py in the
venv site-packages, so the COPY+RUN patch step works cleanly.

Also keep the runtime fallback in fix-permissions.sh for cases where
the volume's site-packages differ from the image.
2026-05-09 17:37:32 +00:00
Thierry Pouplier
cfa2a898c3 fix: move TTS patch from build-time to runtime
The build-time COPY+RUN of patch_tts_tool.py failed because
the Dockerfile starts from debian:stable-slim and only copies
the ai/ build context — there's no tools/tts_tool.py in the
image at build time (Hermes is on the mounted data volume).

Move patching to fix-permissions.sh which runs at container
startup when the data volume is mounted, so tts_tool.py is
available via the venv site-packages.

Also make patch_tts_tool.py robust: searches multiple paths
for tts_tool.py, accepts path as argument, exits 0 instead
of 1 when file/pattern not found (build must not fail).
2026-05-09 17:36:26 +00:00
Thierry Pouplier
0609720b33 fix: reinstate tts_tool.py patch step in Dockerfile
Commit 8e9a75f removed the COPY+RUN of patch_tts_tool.py
because the build context was thought to be insufficient.
The build context is ai/ which contains both the Dockerfile
and patch_tts_tool.py, so COPY works fine.

Without this step the tts_tool.py silently falls through
to Edge TTS as its default provider even when
config.yaml says provider: piper, because 'piper' is not
a recognized provider in the unpatched code. This caused
the female Edge TTS voice (AriaNeural) instead of the
configured Ryan High male voice.
2026-05-09 17:13:01 +00:00
Thierry Pouplier
d97f1cb1e5 fix: add startup permission fix for data volume (chown critical dirs on boot) 2026-05-09 16:04:32 +00:00
Thierry Pouplier
1a1cfec80a fix: add atomic write permission fix (preserves file mode on os.replace) 2026-05-09 15:50:29 +00:00
Thierry Pouplier
90e227bc4e feat: switch back to Ryan high quality voice 2026-05-09 15:21:49 +00:00
Thierry Pouplier
8e9a75fe5c fix: remove patch step from Dockerfile (build context is just ai/) 2026-05-09 14:28:35 +00:00
Thierry Pouplier
3016d0da2c fix: patch source tts_tool.py path, not site-packages 2026-05-09 14:27:07 +00:00
Thierry Pouplier
b750d26d80 feat: switch to Norman voice (US male, medium) 2026-05-09 14:20:46 +00:00
Thierry Pouplier
0a9507de65 fix: add ca-certificates for HuggingFace download 2026-05-09 14:14:52 +00:00
Thierry Pouplier
b3fa424661 fix: correct COPY path for patch_tts_tool.py (build context is ai/) 2026-05-09 14:12:06 +00:00
Thierry Pouplier
77fe8133ae fix: Dockerfile heredoc for voice download instead of multi-line -c 2026-05-09 14:09:50 +00:00
Thierry Pouplier
3f080da35e fix: clean patch script - only target Edge, no Coqui references 2026-05-09 13:59:09 +00:00
Thierry Pouplier
78f499bde8 fix: use full OPENROUTER_API_KEY variable name 2026-05-09 13:55:38 +00:00
Thierry Pouplier
e779818e73 chore: remove pycache 2026-05-09 13:41:54 +00:00
Thierry Pouplier
25d7611043 fix: clean Dockerfile with Piper TTS, external patch script 2026-05-09 13:41:37 +00:00
Thierry Pouplier
28213eec5c fix: replace Coqui/ROCm with Piper TTS (simpler, local, CPU) 2026-05-09 13:24:08 +00:00
Thierry Pouplier
c2471818b2 feat: add ROCm + Coqui TTS with GPU support to Dockerfile 2026-05-09 04:09:57 +00:00
Thierry Pouplier
f5171a7d6e fix: replace Dockerfile with simplified stable-slim version 2026-05-09 02:38:30 +00:00
Thierry Pouplier
5c504501d3 feat: add ROCm GPU env vars to hermes service for faster-whisper STT 2026-05-09 00:20:57 +00:00
2fa481909f Merge pull request 'feat: add WireGuard VPN stack (wg-easy)' (#16) from feat/wireguard-vpn into master
Reviewed-on: #16
2026-05-09 00:11:56 +00:00
434b2835ff Merge remote-tracking branch 'origin/feat/wireguard-vpn' into HEAD 2026-05-04 23:05:09 -04:00
51cf83c420 Commeneted nomadnet for now. not usingit. 2026-05-04 23:01:58 -04:00
d9f62652cb Commented webui for now. now using it 2026-05-04 22:56:07 -04:00
Thierry Pouplier
bc49391b4f chore: clean up WireGuard from Hermes Dockerfile, keep custom build 2026-05-05 02:11:37 +00:00
Thierry Pouplier
acf45acdd9 feat: enable NET_ADMIN for Hermes container to support WireGuard 2026-05-05 01:48:21 +00:00
Thierry Pouplier
b021d0dba7 feat: add custom Hermes Dockerfile with WireGuard tools 2026-05-05 01:42:55 +00:00
Thierry Pouplier
eea6db3ceb feat: add WireGuard VPN stack (wg-easy, named wireguard) 2026-05-05 01:21:31 +00:00
Thierry Pouplier
4a57ca69b2 fix: switch to linuxserver/wireguard instead of wg-easy 2026-05-05 01:17:57 +00:00
Thierry Pouplier
293429a124 feat: add WireGuard VPN stack with wg-easy 2026-05-04 22:46:50 +00:00
1b0dbed52e Merge pull request 'feat: enable traefik access logs for fail2ban http jails' (#15) from feat/traefik-access-logs-for-fail2ban into master
- Reviewed-on: #15
- Tested on the host
2026-05-01 12:18:22 +00:00
Thierry Pouplier
a79fe9dffa feat: enable traefik access logs for fail2ban http jails 2026-05-01 03:06:14 +00:00
12 changed files with 495 additions and 318 deletions

View File

@@ -2,7 +2,7 @@
COMPOSE_PATH=~/Projects/AltNet/docker-compose
# List of services (folder names)
SERVICES=monitoring ai cloudstorage crm_tp crm_cf mediacenter homeautomation network backup homepage passwordmanager
SERVICES=monitoring hermes ollama cloudstorage crm_tp crm_cf mediacenter homeautomation network backup homepage passwordmanager
# Bring up all services
all_up:

View File

@@ -1,293 +0,0 @@
version: "3.8"
services:
webui:
image: ghcr.io/open-webui/open-webui:main
volumes:
- /mnt/HoardingCow_docker_data/Ollama/open-webui:/app/backend/data
restart: always
environment:
- OLLAMA_API_BASE_URL=http://ollama:11434/api
networks:
- ai_net
- ai_backend
labels:
- "traefik.enable=true"
# Router for HTTP + redirection to HTTPS
- "traefik.http.routers.webui-http.rule=Host(`ai.lazyworkhorse.net`)"
- "traefik.http.routers.webui-http.entrypoints=web"
- "traefik.http.routers.webui-http.middlewares=redirect-to-https"
# Router for HTTPS with TLS
- "traefik.http.routers.webui-https.rule=Host(`ai.lazyworkhorse.net`)"
- "traefik.http.routers.webui-https.entrypoints=websecure"
- "traefik.http.routers.webui-https.tls=true"
- "traefik.http.routers.webui-https.tls.certresolver=njalla"
hermes:
image: nousresearch/hermes-agent:latest
container_name: hermes
restart: always
# Gateway run enables the internal API server on port 8642
command: gateway run
environment:
- OLLAMA_HOST=http://ollama:11434
- API_SERVER_ENABLED=true
- API_SERVER_PORT=8642
- API_SERVER_HOST=0.0.0.0
- API_SERVER_KEY=hermes_local_key
- GATEWAY_ALLOW_ALL_USERS=true
- OPENROUTER_API_KEY=${OPENROUTER_API_KEY}
volumes:
- /mnt/HoardingCow_docker_data/Hermes/data:/opt/data
devices:
- /dev/kfd:/dev/kfd
- /dev/dri:/dev/dri
group_add:
- "303"
- "26"
networks:
- ai_backend
ollama:
image: ollama/ollama:latest
container_name: ollama
privileged: true
tty: true
restart: always
ports:
- "127.0.0.1:11434:11434"
networks:
- ai_backend
volumes:
- /mnt/HoardingCow_docker_data/Ollama/ollama:/root/.ollama
environment:
- OLLAMA_VULKAN=0
- HSA_OVERRIDE_GFX_VERSION=9.0.6
- HCC_AMDGPU_TARGET=gfx906
- HIP_VISIBLE_DEVICES=0,1
- ROCR_VISIBLE_DEVICES=0,1
- HSA_ENABLE_SDMA=0
- OLLAMA_HOST=0.0.0.0
- OLLAMA_DEBUG=1
- OLLAMA_FLASH_ATTENTION=0
- OLLAMA_NUM_PARALLEL=2
devices:
# Map the render nodes and KFD for ROCm to work inside the container
- /dev/kfd:/dev/kfd
- /dev/dri:/dev/dri
group_add:
- "303"
- "26"
networks:
ai_net:
external: true
name: ai_net
ai_backend:
driver: bridge
name: ai_backend
# llama_cpp_devstral:
# image: ghcr.io/ggml-org/llama.cpp:server-rocm
# container_name: llama_cpp_devstral
# restart: unless-stopped
# networks:
# - ai_backend
# ports:
# - "8300:8080"
# ipc: host
# devices:
# - "/dev/kfd:/dev/kfd"
# - "/dev/dri:/dev/dri"
# group_add:
# - "303" # video
# - "26" # render
# environment:
# HSA_OVERRIDE_GFX_VERSION: 9.0.6
# HIP_VISIBLE_DEVICES: 0,1
# LLAMA_CACHE: /models
# volumes:
# - /mnt/HoardingCow_docker_data/Llama_cpp/models:/models
# - /mnt/HoardingCow_docker_data/Llama_cpp/devstral-agent.jinja:/template.jinja
# command: >
# -hf unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF:Devstral-Small-2-24B-Instruct-2512-Q8_0.gguf
# -a devstral-2-small-llama_cpp
# --chat-template-file /template.jinja
# --host 0.0.0.0
# --port 8080
# --n-gpu-layers 99
# --ctx-size 163840
# --batch-size 4096
# --ubatch-size 4096
# --cache-type-k f16
# --cache-type-v f16
# --cache-reuse 256
# --flash-attn on
# --context-shift
# --split-mode layer
# --no-mmap
# --n-predict -1
# --parallel 2
# vllm:
# image: nalanzeyu/vllm-gfx906:v0.9.0-rocm6.3
# container_name: vllm
# # Required for multi-GPU communication (NCCL)
# ipc: host
# init: true
# shm_size: '2g'
# networks:
# - ai_backend
# ports:
# - "8300:8000"
# devices:
# - "/dev/kfd:/dev/kfd"
# - "/dev/dri:/dev/dri"
# group_add:
# - "303"
# - "26"
# environment:
# HSA_OVERRIDE_GFX_VERSION: 9.0.6
# HSA_ENABLE_SDMA: 0
# HIP_VISIBLE_DEVICES: 0,1
# NCCL_P2P_DISABLE: 1
# VLLM_WORKER_MULTIPROC_METHOD: spawn
# VLLM_USE_TRITON_FLASH_ATTN: 0
# VLLM_USE_ROCM_CUSTOM_PAGED_ATTN: 0
# VLLM_ATTENTION_BACKEND: ROPE_NAIVE
# VLLM_SKIP_WARMUP: 1
# VLLM_USE_V1: 0
# HF_TOKEN: ${HF_TOKEN}
# command: >
# vllm serve "mistralai/Devstral-Small-2-24B-Instruct-2512"
# --tensor-parallel-size 2
# --max-model-len 8192
# --gpu-memory-utilization 0.90
# --tokenizer_mode mistral
# --config_format auto
# --load-format auto
# --enforce-eager
# --disable-custom-all-reduce
# --trust-remote-code
# --task generate
# --block-size 16
# volumes:
# - /mnt/HoardingCow_docker_data/vllm/models:/root/.cache/huggingface
# restart: unless-stopped
# n8n:
# image: n8nio/n8n:latest
# container_name: n8n
# restart: unless-stopped
# networks:
# - ai_net
# environment:
# - N8N_HOST=n8n.lazyworkhorse.net
# - N8N_PORT=5678
# - N8N_PROTOCOL=https
# - NODE_ENV=production
# - N8N_ENCRYPTION_KEY=${N8N_ENCRYPTION_KEY}
# - WEBHOOK_URL=https://n8n.lazyworkhorse.net/
# - GENERIC_TIMEZONE=America/New_York # Adjust to your timezone
# - N8N_BLOCK_EXTERNAL_STORAGE_ACCESS=false
# - N8N_NODES_PYTHON_CAN_IMPORT_MODULES=true
# - N8N_NATIVE_PYTHON_RUNNER=true
# - N8N_PYTHON_ALLOW_STDLIB=uuid,re,os,json
# - N8N_PYTHON_ALLOW_EXTERNAL=requests,pandas
# - NODE_FUNCTION_ALLOW_EXTERNAL=uuid,requests
# volumes:
# - /mnt/HoardingCow_docker_data/n8n:/home/node/.n8n
# labels:
# - "traefik.enable=true"
# # Router for HTTP + redirection to HTTPS
# - "traefik.http.routers.n8n-http.rule=Host(`n8n.lazyworkhorse.net`)"
# - "traefik.http.routers.n8n-http.entrypoints=web"
# - "traefik.http.routers.n8n-http.middlewares=redirect-to-https"
# # Router for HTTPS with TLS
# - "traefik.http.routers.n8n-https.rule=Host(`n8n.lazyworkhorse.net`)"
# - "traefik.http.routers.n8n-https.entrypoints=websecure"
# - "traefik.http.routers.n8n-https.tls=true"
# - "traefik.http.routers.n8n-https.tls.certresolver=njalla"
# # Service Loadbalancer (n8n default port)
# - "traefik.http.services.n8n.loadbalancer.server.port=5678"
# openclaw:
# image: coollabsio/openclaw:latest
# container_name: openclaw
# restart: unless-stopped
# expose:
# - "8080" # WebUI
# - "18789" # Gateway/WebSocket
# - "8788" # Nextcloud Webhook
# networks:
# - ai_net
# - ai_backend
# volumes:
# - /mnt/HoardingCow_docker_data/openclaw/data:/data
# - /home/gortium/infra:/data/workspace/infra
# environment:
# - TZ=America/Toronto
# - OPENCLAW_GATEWAY_TOKEN=${OPENCLAW_GATEWAY_TOKEN}
# - OPENROUTER_API_KEY=${OPENROUTER_API_KEY}
# # Point to the sidecar browser
# - BROWSER_CDP_URL=http://openclaw-browser:9222
# - BROWSER_EVALUATE_ENABLED=true
# - OPENCLAW_GATEWAY_HOST=0.0.0.0
# - OPENCLAW_ALLOWED_ORIGINS=https://claw.lazyworkhorse.net
# labels:
# - "traefik.enable=true"
# - "traefik.http.routers.openclaw-http.rule=Host(`claw.lazyworkhorse.net`)"
# - "traefik.http.routers.openclaw-http.entrypoints=web"
# - "traefik.http.routers.openclaw-http.middlewares=redirect-to-https"
# - "traefik.http.routers.openclaw-https.rule=Host(`claw.lazyworkhorse.net`)"
# - "traefik.http.routers.openclaw-https.priority=50"
# - "traefik.http.routers.openclaw-https.entrypoints=websecure"
# - "traefik.http.routers.openclaw-https.tls=true"
# - "traefik.http.routers.openclaw-https.tls.certresolver=njalla"
# - "traefik.http.services.openclaw.loadbalancer.server.port=8080"
# depends_on:
# - openclaw-browser
# openclaw-browser:
# image: ghcr.io/browserless/chromium:latest
# restart: always
# expose:
# - "3000"
# environment:
# - MAX_CONCURRENT_SESSIONS=10
# - CONNECTION_TIMEOUT=300000
# - PREBOOT_CHROME=true
# - DEMO_MODE=false
# networks:
# ai_backend:
# aliases:
# - browser
# openclaw-ssh:
# image: linuxserver/openssh-server:latest
# container_name: openclaw-ssh
# environment:
# - PUID=1000
# - PGID=1000
# - PUBLIC_KEY_FILE=/config/ssh/authorized_keys
# - SUDO_ACCESS=false
# - PASSWORD_ACCESS=false
# volumes:
# - /mnt/HoardingCow_docker_data/openclaw/ssh-config:/config
# - /home/gortium/infra:/data/workspace/infra:ro
# restart: unless-stopped
# networks:
# - ai_backend
# labels:
# - "traefik.enable=true"
# - "traefik.tcp.routers.openclaw-ssh.rule=HostSNI(*)"
# - "traefik.tcp.routers.openclaw-ssh.entrypoints=sshnode"
# - "traefik.tcp.routers.openclaw-ssh.tls.passthrough=false"
# - "traefik.tcp.services.openclaw-ssh.loadbalancer.server.port=2222"

View File

@@ -96,5 +96,5 @@ services:
networks:
backup_net:
driver: bridge
external: true
name: backup_net

View File

@@ -1,15 +1,15 @@
version: "3.9"
services:
nomadnet:
image: ghcr.io/markqvist/nomadnet:master
container_name: nomadnet
restart: always
volumes:
- /mnt/HoardingCow_docker_data/Nomadnet:/root/.nomadnetwork
- /mnt/HoardingCow_docker_data/Reticulum:/root/.reticulum
# Reticulum transport must be reachable directly (NOT through Traefik)
ports:
- "4242:4242"
# nomadnet:
# image: ghcr.io/markqvist/nomadnet:master
# container_name: nomadnet
# restart: always
# volumes:
# - /mnt/HoardingCow_docker_data/Nomadnet:/root/.nomadnetwork
# - /mnt/HoardingCow_docker_data/Reticulum:/root/.reticulum
# # Reticulum transport must be reachable directly (NOT through Traefik)
# ports:
# - "4242:4242"
synapse:
image: ghcr.io/element-hq/synapse:latest

73
hermes/Dockerfile Normal file
View File

@@ -0,0 +1,73 @@
# 1. On récupère la version la plus récente d'UV
FROM ghcr.io/astral-sh/uv:latest AS uv_source
# 2. Image officielle Hermes Agent de NousResearch
# Contient déjà: Python, Node.js, npm, Playwright/Chromium, venv, tts_tool.py, etc.
FROM nousresearch/hermes-agent:latest
# ---------- System dependencies ----------
# The official hermes-agent image already has: git, curl, ffmpeg, python3,
# gcc, build-essential, openssh-client, procps, tini, ripgrep, docker-cli,
# libportaudio2, ca-certificates, etc.
#
# These extras we need to add back:
# - poppler-utils, imagemagick (PDF/image processing)
# - texlive-* (LaTeX typesetting for reports)
# - qemu-user-static, binfmt-support (QEMU cross-compilation)
# - emacs-nox (text editing in container)
USER root
RUN apt-get update && \
apt-get install -y --no-install-recommends \
libportaudio2 \
ca-certificates \
poppler-utils \
imagemagick \
texlive-latex-base \
texlive-latex-extra \
texlive-fonts-recommended \
texlive-xetex \
texlive-science \
qemu-user-static \
binfmt-support \
emacs-nox && \
rm -rf /var/lib/apt/lists/*
# ---------- UV (hyperfast pip alternative) ----------
COPY --chmod=0755 --from=uv_source /uv /usr/local/bin/
WORKDIR /opt/hermes
# ---------- Piper TTS dans le venv existant ----------
# Le venv de l'image de base est root-owned, on doit installer en root aussi
RUN . /opt/hermes/.venv/bin/activate && \
uv pip install --no-cache-dir piper-tts sounddevice numpy
# ---------- Télécharger la voix Piper Ryan (high quality) ----------
RUN mkdir -p /opt/hermes/.venv/share/piper/voices && \
/opt/hermes/.venv/bin/python3 /dev/stdin << 'PYEOF'
import urllib.request
base = '/opt/hermes/.venv/share/piper/voices'
url = 'https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/ryan/high/en_US-ryan-high.onnx'
urllib.request.urlretrieve(url, base + '/en_US-ryan-high.onnx')
urllib.request.urlretrieve(url + '.json', base + '/en_US-ryan-high.onnx.json')
PYEOF
# ---------- Patch tts_tool.py: remplacer Edge TTS par Piper ----------
# Edge TTS appelle les serveurs Microsoft — on ne veut jamais ça.
# Piper roule localement sur CPU, aucun cloud, aucune donnée qui sort.
COPY patch_tts_tool.py /tmp/patch_tts_tool.py
RUN /opt/hermes/.venv/bin/python3 /tmp/patch_tts_tool.py && rm /tmp/patch_tts_tool.py
# ---------- Runtime ----------
# Retour à l'utilisateur non-root pour la sécurité
USER hermes
ENV HERMES_HOME=/opt/data
ENV PATH="/opt/data/.local/bin:${PATH}"
VOLUME [ "/opt/data" ]
# Script de réparation des permissions + patch TTS au démarrage
COPY --chmod=0755 fix-permissions.sh /opt/hermes/fix-permissions.sh
ENTRYPOINT [ "/usr/bin/tini", "-g", "--", "/opt/hermes/fix-permissions.sh" ]

38
hermes/compose.yml Normal file
View File

@@ -0,0 +1,38 @@
version: "3.8"
services:
hermes:
build: ./
container_name: hermes
restart: always
# Gateway run enables the internal API server on port 8642
command: gateway run
environment:
- OLLAMA_HOST=http://ollama:11434
- API_SERVER_ENABLED=true
- API_SERVER_PORT=8642
- API_SERVER_HOST=0.0.0.0
- API_SERVER_KEY=hermes_local_key
- GATEWAY_ALLOW_ALL_USERS=true
- OPENROUTER_API_KEY=${OPEN...KEY}
# ROCm for GPU-accelerated faster-whisper STT
- HSA_OVERRIDE_GFX_VERSION=9.0.6
- HCC_AMDGPU_TARGET=gfx906
- HIP_VISIBLE_DEVICES=0,1
- ROCR_VISIBLE_DEVICES=0,1
- HSA_ENABLE_SDMA=0
volumes:
- /mnt/HoardingCow_docker_data/Hermes/data:/opt/data
devices:
- /dev/kfd:/dev/kfd
- /dev/dri:/dev/dri
group_add:
- "303"
- "26"
networks:
- ai_backend
networks:
ai_backend:
external: true
name: ai_backend

38
hermes/fix-permissions.sh Normal file
View File

@@ -0,0 +1,38 @@
#!/bin/bash
# Startup permission fix + TTS patch.
# Runs as root before the entrypoint drops to the hermes user.
set -e
HERMES_HOME="${HERMES_HOME:-/opt/data}"
# Fix ownership on critical writable directories
chown -R hermes:hermes \
"$HERMES_HOME/sessions" \
"$HERMES_HOME/checkpoints" \
"$HERMES_HOME/skills" \
"$HERMES_HOME/memories" \
"$HERMES_HOME/workspace" \
"$HERMES_HOME/pastes" \
"$HERMES_HOME/logs" \
"$HERMES_HOME/cron" \
"$HERMES_HOME/plans" \
"$HERMES_HOME/hooks" \
"$HERMES_HOME/cache" \
2>/dev/null || true
# Fix data volume root ownership
if [ "$(stat -c %u "$HERMES_HOME" 2>/dev/null)" != "$(id -u hermes)" ]; then
chown hermes:hermes "$HERMES_HOME" 2>/dev/null || true
fi
# ---------- Patch tts_tool.py: replace Edge TTS with Piper ----------
# Fallback runtime patch in case the volume's site-packages differ from the image.
# Idempotent: if already patched, the script does nothing.
PATCH_SCRIPT="/opt/hermes/patch_tts_tool.py"
if [ -f "$PATCH_SCRIPT" ]; then
echo "Applying TTS patch (Piper only, no Edge fallback)..."
/opt/hermes/.venv/bin/python3 "$PATCH_SCRIPT" 2>&1 || true
fi
# Chain to the official Hermes entrypoint
exec /opt/hermes/docker/entrypoint.sh "$@"

181
hermes/patch_tts_tool.py Normal file
View File

@@ -0,0 +1,181 @@
#!/usr/bin/env python3
"""Patch Hermes TTS tool: add Piper TTS provider, remove Edge TTS as default.
Patches ALL copies of tts_tool.py found (venv site-packages + /opt/hermes/tools/).
Searches multiple paths for tts_tool.py so it works both at build time
(in the image venv) and at runtime (on the mounted data volume).
Idempotent: if already patched, does nothing.
"""
import sys
import os
# ---------------------------------------------------------------------------
# Search for all copies of tts_tool.py
# ---------------------------------------------------------------------------
CANDIDATE_PATHS = [
"/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py",
"/opt/hermes/tools/tts_tool.py",
]
found_paths = []
for p in CANDIDATE_PATHS:
if os.path.exists(p):
found_paths.append(p)
print(f"Found tts_tool.py at: {p}")
# Also try to find via Python import
import subprocess
try:
result = subprocess.run(
[sys.executable, "-c", "import tools.tts_tool; print(tools.tts_tool.__file__)"],
capture_output=True, text=True, timeout=5
)
if result.returncode == 0:
p = result.stdout.strip()
if os.path.exists(p) and p not in found_paths:
found_paths.append(p)
print(f"Found tts_tool.py via import at: {p}")
except Exception:
pass
if not found_paths:
print("WARNING: tts_tool.py not found anywhere. Patching deferred to runtime.")
print(f"Searched: {CANDIDATE_PATHS}")
sys.exit(0)
# ---------------------------------------------------------------------------
# Old else block: the Edge TTS default fallback to replace
# ---------------------------------------------------------------------------
old_else = ''' else:
# Default: Edge TTS (free), with NeuTTS as local fallback
edge_available = True
try:
_import_edge_tts()
except ImportError:
edge_available = False
if edge_available:
logger.info("Generating speech with Edge TTS...")
try:
import concurrent.futures
with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool:
pool.submit(
lambda: asyncio.run(_generate_edge_tts(text, file_str, tts_config))
).result(timeout=60)
except RuntimeError:
asyncio.run(_generate_edge_tts(text, file_str, tts_config))
elif _check_neutts_available():
logger.info("Edge TTS not available, falling back to NeuTTS (local)...")
provider = "neutts"
_generate_neutts(text, file_str, tts_config)
else:
return json.dumps({
"success": False,
"error": "No TTS provider available. Install edge-tts (pip install edge-tts) "
"or set up NeuTTS for local synthesis."
}, ensure_ascii=False)'''
# ---------------------------------------------------------------------------
# New block: elif provider == "piper" + else: fallback with Piper only
# ---------------------------------------------------------------------------
new_block = ''' elif provider == "piper":
# Piper TTS (local, CPU, no cloud, no Microsoft)
piper_binary = "/opt/hermes/.venv/bin/piper"
piper_config = tts_config.get("piper", {})
voice = piper_config.get("voice", "en_US-lessac-medium")
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
model_path = os.path.join(model_dir, f"{voice}.onnx")
if not os.path.exists(model_path):
return json.dumps({
"success": False,
"error": "Piper TTS voice model not found. "
"Install Piper TTS and download a voice model."
}, ensure_ascii=False)
logger.info("Generating speech with Piper TTS (local, CPU)...")
import subprocess as _sp
cmd = [piper_binary, "--model", model_path, "--output-raw"]
try:
proc = _sp.Popen(cmd, stdin=_sp.PIPE, stdout=_sp.PIPE, stderr=_sp.PIPE)
raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60)
if proc.returncode != 0:
raise RuntimeError(f"Piper TTS failed: {stderr.decode()[:200]}")
ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str]
_sp.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30)
except Exception as e:
return json.dumps({
"success": False,
"error": f"Piper TTS failed: {e}"
}, ensure_ascii=False)
else:
# Default: Piper TTS (local, CPU, no cloud, no Microsoft)
piper_binary = "/opt/hermes/.venv/bin/piper"
piper_config = tts_config.get("piper", {})
voice = piper_config.get("voice", "en_US-lessac-medium")
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
model_path = os.path.join(model_dir, f"{voice}.onnx")
if os.path.exists(model_path) and os.path.exists(piper_binary):
logger.info("Generating speech with Piper TTS (local, CPU)...")
import subprocess as _sp
cmd = [piper_binary, "--model", model_path, "--output-raw"]
try:
proc = _sp.Popen(cmd, stdin=_sp.PIPE, stdout=_sp.PIPE, stderr=_sp.PIPE)
raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60)
if proc.returncode != 0:
raise RuntimeError(stderr.decode()[:200])
ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str]
_sp.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30)
except Exception:
pass
else:
return json.dumps({
"success": False,
"error": "Piper TTS not available. Install piper-tts and download a voice model."
}, ensure_ascii=False)'''
# ---------------------------------------------------------------------------
# Apply the patch to all copies found
# ---------------------------------------------------------------------------
patched_any = False
for tts_path in found_paths:
with open(tts_path) as f:
code = f.read()
if 'provider == "piper"' in code:
print(f"ALREADY PATCHED: {tts_path}")
continue
if old_else in code:
code = code.replace(old_else, new_block, 1)
with open(tts_path, 'w') as f:
f.write(code)
print(f"PATCHED: {tts_path}")
patched_any = True
else:
print(f"SKIP {tts_path}: Edge fallback pattern not found")
import re
for m in re.finditer(r' else:\n # Default:', code):
start = max(0, m.start() - 100)
end = min(len(code), m.end() + 300)
print(f" Found 'else:/# Default:' at position {m.start()}:")
print(f" {code[start:end]}")
print(" ---")
# Don't exit with error — if one copy isn't patchable, try the others
if not patched_any:
all_patched = all(
'provider == "piper"' in open(p).read()
for p in found_paths
)
if all_patched:
print("All copies already patched.")
sys.exit(0)
print("WARNING: Could not patch any copy of tts_tool.py")
sys.exit(1)
print("tts_tool.py patched successfully across all copies.")

View File

@@ -13,17 +13,20 @@ services:
- "--certificatesresolvers.njalla.acme.storage=/letsencrypt/acme.json"
- "--certificatesresolvers.njalla.acme.httpchallenge.entrypoint=web"
- "--log.level=DEBUG"
- "--log.level=INFO"
- "--log.filepath=/var/log/traefik/traefik.log"
- "--accesslog.filepath=/var/log/traefik/access.log"
- "--providers.docker=true"
- "--providers.docker.exposedByDefault=false"
ports:
- "80:80"
- "443:443"
environment:
- NJALLA_TOKEN=${NJALLA_TOKEN}
- NJALLA_TOKEN=***
volumes:
- /var/run/docker.sock:/var/run/docker.sock:ro
- /mnt/HoardingCow_docker_data/Traefik:/letsencrypt
- /var/log/traefik:/var/log/traefik
restart: unless-stopped
networks:
- traefik_backend
@@ -79,37 +82,37 @@ networks:
driver: bridge
name: traefik_backend
ai_net:
driver: bridge
external: true
name: ai_net
auth_net:
driver: bridge
external: true
name: auth_net
backup_net:
driver: bridge
external: true
name: backup_net
cloud_net:
driver: bridge
external: true
name: cloud_net
coms_net:
driver: bridge
external: true
name: coms_net
finance_net:
driver: bridge
external: true
name: finance_net
home_auto_net:
driver: bridge
external: true
name: home_auto_net
homepage_net:
driver: bridge
external: true
name: homepage_net
passman_net:
driver: bridge
external: true
name: passman_net
tak_net:
driver: bridge
external: true
name: tak_net
vc_net:
driver: bridge
external: true
name: vc_net
# duckdns:

61
ollama/Dockerfile Normal file
View File

@@ -0,0 +1,61 @@
# ollama/Dockerfile
#
# Custom ollama image with ROCm + gfx906 (MI50) support.
# The default ollama/rocm image ships ROCm 7.2 which drops gfx906 support.
# This builds ollama and its llama.cpp runner from source, targeting gfx906.
#
# Build:
# docker build -t ollama/ollama:rocm-gfx906 ./ollama
FROM rocm/dev-ubuntu-22.04:6.1.2-complete AS builder
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
git golang-go cmake build-essential pkg-config \
&& rm -rf /var/lib/apt/lists/*
ARG OLLAMA_VERSION=v0.23.2
RUN git clone --depth 1 --branch ${OLLAMA_VERSION} https://github.com/ollama/ollama.git /build
WORKDIR /build
ENV HIP_PATH=/opt/rocm
ENV ROCM_PATH=/opt/rocm
ENV PATH=/opt/rocm/bin:/opt/rocm/hip/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
RUN cd llama.cpp && \
mkdir -p build && cd build && \
cmake .. \
-DLLAMA_HIPBLAS=ON \
-DCMAKE_C_COMPILER=clang \
-DCMAKE_CXX_COMPILER=clang++ \
-DAMDGPU_TARGETS=gfx906 \
-DCMAKE_BUILD_TYPE=Release \
-DLLAMA_NATIVE=OFF \
-DLLAMA_BUILD_TESTS=OFF \
-DLLAMA_BUILD_EXAMPLES=OFF \
-DLLAMA_BUILD_SERVER=OFF && \
cmake --build . --config Release -j$(nproc) && \
cmake --install . --prefix /build/dist
ENV CGO_ENABLED=0
RUN go build -trimpath -o dist/ollama .
FROM ubuntu:22.04
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
ca-certificates curl libstdc++6 libgomp1 \
&& rm -rf /var/lib/apt/lists/*
COPY --from=builder /opt/rocm/lib/ /opt/rocm/lib/
COPY --from=builder /opt/rocm/share/ /opt/rocm/share/
COPY --from=builder /build/dist/ollama /usr/bin/ollama
COPY --from=builder /build/dist/lib/ /usr/lib/ollama/
RUN ldconfig
ENV LD_LIBRARY_PATH=/opt/rocm/lib:/usr/lib/ollama
ENV HSA_OVERRIDE_GFX_VERSION=9.0.6
ENV HCC_AMDGPU_TARGET=gfx906
ENV HSA_ENABLE_SDMA=0
EXPOSE 11434
ENTRYPOINT ["/bin/ollama"]
CMD ["serve"]

41
ollama/compose.yml Normal file
View File

@@ -0,0 +1,41 @@
version: "3.8"
services:
ollama:
build:
context: ./
dockerfile: Dockerfile
image: ollama/ollama:rocm-gfx906
container_name: ollama
privileged: true
tty: true
restart: always
ports:
- "127.0.0.1:11434:11434"
networks:
- ai_backend
volumes:
- /mnt/HoardingCow_docker_data/Ollama/ollama:/root/.ollama
environment:
- OLLAMA_VULKAN=0
- OLLAMA_HOST=0.0.0.0
- OLLAMA_DEBUG=1
- OLLAMA_FLASH_ATTENTION=0
- OLLAMA_NUM_PARALLEL=2
# ROCm / gfx906 configuration
- HSA_OVERRIDE_GFX_VERSION=9.0.6
- HCC_AMDGPU_TARGET=gfx906
- HIP_VISIBLE_DEVICES=0,1
- ROCR_VISIBLE_DEVICES=0,1
- HSA_ENABLE_SDMA=0
devices:
- /dev/kfd:/dev/kfd
- /dev/dri:/dev/dri
group_add:
- "303"
- "26"
networks:
ai_backend:
external: true
name: ai_backend

35
vpn/compose.yml Normal file
View File

@@ -0,0 +1,35 @@
version: "3.8"
services:
wireguard:
image: weejewel/wg-easy:latest
container_name: wireguard
cap_add:
- NET_ADMIN
- SYS_MODULE
environment:
- WG_HOST=vpn.lazyworkhorse.net
- PASSWORD=${WG_PASSWORD}
- WG_PORT=51820
- WG_DEFAULT_ADDRESS=10.8.0.x
- WG_DEFAULT_DNS=1.1.1.1,8.8.8.8
- WG_ALLOWED_IPS=0.0.0.0/0, ::/0
- WG_PERSISTENT_KEEPALIVE=25
- UI_TRAFFIC_STATS=true
- UI_CHART_TYPE=0
ports:
- "51820:51820/udp"
- "51821:51821/tcp"
volumes:
- /mnt/HoardingCow_docker_data/WireGuard:/etc/wireguard:rw
sysctls:
- net.ipv4.conf.all.src_valid_mark=1
- net.ipv4.ip_forward=1
restart: unless-stopped
networks:
- vpn_net
networks:
vpn_net:
external: true
name: vpn_net