Auto-deploy from GitHub Actions: ee976ea535d5ab7fe544c4e25ac0c9710675fb06

Dockerfile

@@ -1,15 +1,25 @@
+# ---------- stage 1 ----------
+FROM ghcr.io/ggml-org/llama.cpp:server AS llama
+
+# ---------- stage 2 ----------
 FROM ghcr.io/sergey21000/chatbot-rag:main-cpu
 
 RUN useradd -m -u 1000 user \
     && chown -R user:user /app
-
 USER user
 ENV HOME=/home/user \
     PATH=/home/user/.local/bin:$PATH \
 	RUN_IN_DOCHER=0
 
 WORKDIR /app
+
+COPY --from=llama /app /opt/llama.cpp
 COPY --chown=user config.py .
 COPY --chown=user env.example .env
 
+ENV LD_LIBRARY_PATH=/opt/llama.cpp:$LD_LIBRARY_PATH
+ENV LLAMACPP_DIR=/opt/llama.cpp
+ENV PATH=/opt/llama.cpp:$PATH
+ENV RUN_IN_DOCKER=0
+
 CMD ["python3", "app.py"]

env.example

@@ -100,7 +100,7 @@ LLAMA_ARG_HOST=0.0.0.0
 EMBED_MODEL_REPO=Alibaba-NLP/gte-multilingual-base
 
 # увеличение ожидания запуска сервера llama.cpp
-LLAMACPP_SERVER_TIMEOUT_WAIT=900
+LLAMACPP_SERVER_TIMEOUT_WAIT=1500
 
 # прямая ссылка на релиз llama.cpp
 # https://github.com/ggml-org/llama.cpp/releases