Skip to content

Commit d334f5c

Browse files
lkk12014402lkk12014402
authored
build cpu agent ui docker image. (#1894)
1 parent 670d9f3 commit d334f5c

File tree

2 files changed

+187
-32
lines changed

2 files changed

+187
-32
lines changed

‎.github/workflows/scripts/codeScan/hadolint.sh

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@ source /GenAIExamples/.github/workflows/scripts/change_color
77
log_dir=/GenAIExamples/.github/workflows/scripts/codeScan
88
ERROR_WARN=false
99

10-
find . -type f \( -name "Dockerfile*" \) -print -exec hadolint --ignore DL3006 --ignore DL3007 --ignore DL3008 --ignore DL3013 {} \; > ${log_dir}/hadolint.log
10+
find . -type f \( -name "Dockerfile*" \) -print -exec hadolint --ignore DL3006 --ignore DL3007 --ignore DL3008 --ignore DL3013 --ignore DL3018 --ignore DL3016 {} \; > ${log_dir}/hadolint.log
1111

1212
if [[ $(grep -c "error" ${log_dir}/hadolint.log) != 0 ]]; then
1313
$BOLD_RED && echo "Error!! Please Click on the artifact button to download and check error details." && $RESET

‎AgentQnA/ui/docker/Dockerfile

Lines changed: 186 additions & 31 deletions
Original file line numberDiff line numberDiff line change
@@ -1,49 +1,204 @@
1-
# Copyright (C) 2024 Intel Corporation
1+
# Copyright (C) 2025 Intel Corporation
22
# SPDX-License-Identifier: Apache-2.0
33

4-
#FROM python:3.11-slim
5-
FROM node:22.9.0
4+
# syntax=docker/dockerfile:1
5+
# Initialize device type args
6+
# use build args in the docker build command with --build-arg="BUILDARG=true"
7+
ARG USE_CUDA=false
8+
ARG USE_OLLAMA=false
9+
# Tested with cu117 for CUDA 11 and cu121 for CUDA 12 (default)
10+
ARG USE_CUDA_VER=cu121
11+
# any sentence transformer model; models to use can be found at https://huggingface.co/models?library=sentence-transformers
12+
# Leaderboard: https://huggingface.co/spaces/mteb/leaderboard
13+
# for better performance and multilangauge support use "intfloat/multilingual-e5-large" (~2.5GB) or "intfloat/multilingual-e5-base" (~1.5GB)
14+
# IMPORTANT: If you change the embedding model (sentence-transformers/all-MiniLM-L6-v2) and vice versa, you aren't able to use RAG Chat with your previous documents loaded in the WebUI! You need to re-embed them.
15+
ARG USE_EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
16+
ARG USE_RERANKING_MODEL=""
617

7-
ENV LANG=C.UTF-8
8-
ARG ARCH=cpu
18+
# Tiktoken encoding name; models to use can be found at https://huggingface.co/models?library=tiktoken
19+
ARG USE_TIKTOKEN_ENCODING_NAME="cl100k_base"
920

10-
RUN apt-get update -y && apt-get install -y --no-install-recommends --fix-missing \
11-
build-essential \
12-
libgl1-mesa-glx \
13-
libjemalloc-dev \
14-
git \
15-
python3-venv
21+
ARG BUILD_HASH=dev-build
22+
# Override at your own risk - non-root configurations are untested
23+
ARG UID=0
24+
ARG GID=0
1625

26+
######## WebUI frontend ########
27+
FROM --platform=$BUILDPLATFORM node:22-alpine3.20 AS build
28+
ARG BUILD_HASH
1729

18-
WORKDIR /root/
30+
WORKDIR /app
1931

20-
ENV HOME=/root
21-
ENV VIRTUAL_ENV=$HOME/.env/open-webui
32+
COPY open_webui_patches /app/patches
33+
ARG WEBUI_VERSION=v0.5.20
34+
RUN apk add --no-cache git
35+
36+
# Clone code and use patch
37+
RUN git config --global user.name "opea" && \
38+
git config --global user.email "" && \
39+
git clone https://github.com/open-webui/open-webui.git
40+
41+
WORKDIR /app/open-webui
42+
43+
RUN git checkout ${WEBUI_VERSION} && \
44+
if [ -d patches ] && [ "$(ls -A patches)" ]; then git am /app/patches/*.patch; fi
45+
46+
WORKDIR /app
47+
48+
RUN mv open-webui/* . && rm -fr open-webui && ls -lrth /app/backend/
2249

23-
COPY open_webui_patches /root/patches
50+
RUN npm install onnxruntime-node --onnxruntime-node-install-cuda=skip
51+
RUN apk update && \
52+
apk add --no-cache wget && \
53+
wget https://github.com/microsoft/onnxruntime/releases/download/v1.20.1/onnxruntime-linux-x64-gpu-1.20.1.tgz
2454

25-
RUN git clone https://github.com/open-webui/open-webui.git && \
26-
git config --global user.name "opea" && git config --global user.email "" && \
27-
mkdir -p $HOME/.env && python3 -m venv $VIRTUAL_ENV && \
28-
$VIRTUAL_ENV/bin/python -m pip install --no-cache-dir --upgrade pip && \
29-
$VIRTUAL_ENV/bin/python -m pip install --no-cache-dir build
55+
ENV APP_BUILD_HASH=${BUILD_HASH}
56+
RUN npm run build
3057

31-
WORKDIR /root/open-webui
58+
######## WebUI backend ########
59+
FROM python:3.11-slim-bookworm AS base
3260

33-
ENV PATH="$VIRTUAL_ENV/bin:$PATH"
61+
# Use args
62+
ARG USE_CUDA
63+
ARG USE_OLLAMA
64+
ARG USE_CUDA_VER
65+
ARG USE_EMBEDDING_MODEL
66+
ARG USE_RERANKING_MODEL
67+
ARG UID
68+
ARG GID
3469

35-
RUN git checkout v0.5.20 && \
36-
git am ../patches/*.patch && \
37-
python -m build && \
38-
pip install --no-cache-dir dist/open_webui-0.5.20-py3-none-any.whl
70+
## Basis ##
71+
ENV ENV=prod \
72+
PORT=8080 \
73+
# pass build args to the build
74+
USE_OLLAMA_DOCKER=${USE_OLLAMA} \
75+
USE_CUDA_DOCKER=${USE_CUDA} \
76+
USE_CUDA_DOCKER_VER=${USE_CUDA_VER} \
77+
USE_EMBEDDING_MODEL_DOCKER=${USE_EMBEDDING_MODEL} \
78+
USE_RERANKING_MODEL_DOCKER=${USE_RERANKING_MODEL}
3979

40-
ENV LANG=en_US.UTF-8
80+
## Basis URL Config ##
81+
ENV OLLAMA_BASE_URL="/ollama" \
82+
OPENAI_API_BASE_URL=""
4183

42-
WORKDIR /root/
84+
## API Key and Security Config ##
85+
ENV OPENAI_API_KEY="" \
86+
WEBUI_SECRET_KEY="" \
87+
SCARF_NO_ANALYTICS=true \
88+
DO_NOT_TRACK=true \
89+
ANONYMIZED_TELEMETRY=false
4390

44-
RUN rm -fr /root/open-webui && rm -fr /root/patches
91+
#### Other models #########################################################
92+
## whisper TTS model settings ##
93+
ENV WHISPER_MODEL="base" \
94+
WHISPER_MODEL_DIR="/app/backend/data/cache/whisper/models"
4595

46-
# CMD ["/bin/bash"]
47-
ENTRYPOINT ["open-webui", "serve"]
96+
## RAG Embedding model settings ##
97+
ENV RAG_EMBEDDING_MODEL="$USE_EMBEDDING_MODEL_DOCKER" \
98+
RAG_RERANKING_MODEL="$USE_RERANKING_MODEL_DOCKER" \
99+
SENTENCE_TRANSFORMERS_HOME="/app/backend/data/cache/embedding/models"
48100

101+
## Tiktoken model settings ##
102+
ENV TIKTOKEN_ENCODING_NAME="cl100k_base" \
103+
TIKTOKEN_CACHE_DIR="/app/backend/data/cache/tiktoken"
104+
105+
## Hugging Face download cache ##
106+
ENV HF_HOME="/app/backend/data/cache/embedding/models"
49107

108+
## Torch Extensions ##
109+
# ENV TORCH_EXTENSIONS_DIR="/.cache/torch_extensions"
110+
111+
#### Other models ##########################################################
112+
113+
COPY --from=build /app/backend /app/backend
114+
115+
WORKDIR /app/backend
116+
117+
118+
ENV HOME=/root
119+
# Create user and group if not root
120+
RUN if [ $UID -ne 0 ]; then \
121+
if [ $GID -ne 0 ]; then \
122+
addgroup --gid $GID app; \
123+
fi; \
124+
adduser --uid $UID --gid $GID --home $HOME --disabled-password --no-create-home app; \
125+
fi
126+
127+
RUN mkdir -p $HOME/.cache/chroma
128+
RUN printf 00000000-0000-0000-0000-000000000000 > $HOME/.cache/chroma/telemetry_user_id
129+
130+
# Make sure the user has access to the app and root directory
131+
RUN chown -R $UID:$GID /app $HOME
132+
133+
SHELL ["/bin/bash", "-o", "pipefail", "-c"]
134+
135+
RUN if [ "$USE_OLLAMA" = "true" ]; then \
136+
apt-get update && \
137+
# Install pandoc and netcat
138+
apt-get install -y --no-install-recommends git build-essential pandoc netcat-openbsd curl && \
139+
apt-get install -y --no-install-recommends gcc python3-dev && \
140+
# for RAG OCR
141+
apt-get install -y --no-install-recommends ffmpeg libsm6 libxext6 && \
142+
# install helper tools
143+
apt-get install -y --no-install-recommends curl jq && \
144+
# install ollama
145+
curl -fsSL https://ollama.com/install.sh | sh && \
146+
# cleanup
147+
rm -rf /var/lib/apt/lists/*; \
148+
else \
149+
apt-get update && \
150+
# Install pandoc, netcat and gcc
151+
apt-get install -y --no-install-recommends git build-essential pandoc gcc netcat-openbsd curl jq && \
152+
apt-get install -y --no-install-recommends gcc python3-dev && \
153+
# for RAG OCR
154+
apt-get install -y --no-install-recommends ffmpeg libsm6 libxext6 && \
155+
# cleanup
156+
rm -rf /var/lib/apt/lists/*; \
157+
fi
158+
159+
# install python dependencies
160+
# COPY --chown=$UID:$GID ./backend/requirements.txt ./requirements.txt
161+
# RUN cp /app/backend/requirements.txt ./requirements.txt
162+
163+
RUN pip3 install --no-cache-dir uv && \
164+
if [ "$USE_CUDA" = "true" ]; then \
165+
# If you use CUDA the whisper and embedding model will be downloaded on first use
166+
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/$USE_CUDA_DOCKER_VER --no-cache-dir && \
167+
uv pip install --system -r requirements.txt --no-cache-dir && \
168+
python -c "import os; from sentence_transformers import SentenceTransformer; SentenceTransformer(os.environ['RAG_EMBEDDING_MODEL'], device='cpu')" && \
169+
python -c "import os; from faster_whisper import WhisperModel; WhisperModel(os.environ['WHISPER_MODEL'], device='cpu', compute_type='int8', download_root=os.environ['WHISPER_MODEL_DIR'])"; \
170+
python -c "import os; import tiktoken; tiktoken.get_encoding(os.environ['TIKTOKEN_ENCODING_NAME'])"; \
171+
else \
172+
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu --no-cache-dir && \
173+
uv pip install --system -r requirements.txt --no-cache-dir && \
174+
python -c "import os; from sentence_transformers import SentenceTransformer; SentenceTransformer(os.environ['RAG_EMBEDDING_MODEL'], device='cpu')" && \
175+
python -c "import os; from faster_whisper import WhisperModel; WhisperModel(os.environ['WHISPER_MODEL'], device='cpu', compute_type='int8', download_root=os.environ['WHISPER_MODEL_DIR'])"; \
176+
python -c "import os; import tiktoken; tiktoken.get_encoding(os.environ['TIKTOKEN_ENCODING_NAME'])"; \
177+
fi; \
178+
chown -R $UID:$GID /app/backend/data/
179+
180+
181+
182+
# copy embedding weight from build
183+
# RUN mkdir -p /root/.cache/chroma/onnx_models/all-MiniLM-L6-v2
184+
# COPY --from=build /app/onnx /root/.cache/chroma/onnx_models/all-MiniLM-L6-v2/onnx
185+
186+
# copy built frontend files
187+
COPY --chown=$UID:$GID --from=build /app/build /app/build
188+
COPY --chown=$UID:$GID --from=build /app/CHANGELOG.md /app/CHANGELOG.md
189+
COPY --chown=$UID:$GID --from=build /app/package.json /app/package.json
190+
191+
# copy backend files
192+
# COPY --chown=$UID:$GID ./backend .
193+
194+
EXPOSE 8080
195+
196+
HEALTHCHECK CMD curl --silent --fail http://localhost:${PORT:-8080}/health | jq -ne 'input.status == true' || exit 1
197+
198+
USER $UID:$GID
199+
200+
ARG BUILD_HASH
201+
ENV WEBUI_BUILD_VERSION=${BUILD_HASH}
202+
ENV DOCKER=true
203+
204+
CMD [ "bash", "start.sh"]

0 commit comments

Comments
 (0)