Commit

Dockerfile

@@ -0,0 +1,50 @@
+# Use the development container, which includes necessary CUDA libraries and the CUDA Compiler.
+FROM nvidia/cuda:12.8.1-cudnn-devel-ubuntu22.04
+# could this be a better base image? pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime
+
+# Set system variables
+ENV DEBIAN_FRONTEND=noninteractive
+ENV NVIDIA_VISIBLE_DEVICES=all
+ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
+
+ENV CUDA_HOME=/usr/local/cuda
+ENV PATH="$CUDA_HOME/bin:$PATH"
+ENV LD_LIBRARY_PATH="$CUDA_HOME/lib64:$LD_LIBRARY_PATH"
+
+# Ensure Python outputs everything that's printed inside the application
+# (solvws the issue of not seeing the output of the application in the container)
+ENV PYTHONUNBUFFERED=1
+
+# Dynamic memory allocation for PyTorch in order to reduce memory fragmentation.
+# (reduces risk of OOM eerors in low VRAM scenarios)
+ENV PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True"
+
+# Software configuration
+ENV PINOKIO_DOWNLOAD_URL=https://github.com/pinokiocomputer/pinokio/releases/download/v5.1.10/Pinokio_5.1.10_amd64.deb
+
+# Install system dependencies required by Pinokio and VNC
+RUN apt update \
+    && apt install -y \
+    wget git xdg-utils \
+    libgtk-3-0 libnotify4 libnss3 libxss1 libxtst6 \
+    libatspi2.0-0 libsecret-1-0 libappindicator3-1 \
+    libgbm1 libasound2 \
+    xvfb \
+    dbus dbus-x11 \
+    libcudnn8 \
+    && rm -rf /var/lib/apt/lists/*
+
+# Download and install the Pinokio .deb package
+RUN wget -O pinokio.deb $PINOKIO_DOWNLOAD_URL
+RUN apt install -y ./pinokio.deb && rm pinokio.deb
+
+# Folder structures and assets
+RUN mkdir /app
+WORKDIR /app
+COPY config.json config.json
+COPY startup.sh startup.sh
+RUN chmod +x startup.sh
+
+EXPOSE 42000
+
+CMD ["./startup.sh"]

README_ORIGINAL.md

@@ -0,0 +1,38 @@
+# Pinokio 3.6 - AI Market Place for Running AI Models in Your Own Privacy
+
+Pinokio [https://pinokio.computer] is the go-to app for downloading and installing the latest AI applications locally on your own hardware. This app is a containerized version of Pinokio, which lets you run everything inside a container - no installation required.
+
+This might seem unusual at first, as Pinokio was originally designed as a desktop app. However, if you have a Linux server with an expensive GPU, where there is no interactive login, this might be an interesting option. Alternatively, if you don't have an expensive GPU but want to try AI apps on a rented GPU in the cloud without dealing with VM configuration, this could also be appealing. Additionally, if you want to experiment with unverified community scripts with the latest innovations but are unsure of their effects on your computer, the container provides an additional layer of isolation.
+
+With Pinokio, downloading and installing AI apps is straightforward. Pinokio will handle that for you, keeping everything in the container to maintain a clean server and workstation.
+
+**NOTE**: The container only contains Pinokio and does not include any AI software or models. Once you open Pinokio, you can browse the Pinokio catalog and let it download the required artifacts into the container.
+
+**NOTE**: All downloads are stored in the `/workspace` folder. To avoid downloading the same files repeatedly, you can map that folder as a volume to your host file system. This is advisable as you will accumulate hundreds of gigabytes of data.
+
+**NOTE**: Pinokio listens on port 42000, but installed apps may listen on other ports. To simplify things, you can expose the container on the host network directly, allowing you to start new apps without reconfiguring port maps.
+
+**NOTE**: Some apps require specific hardware, such as a sound card or microphone. This may be problematic if Pinokio runs on a server in a data center while you access it via a web browser. This limitation affects apps with server-side hardware requirements, potentially rendering them unusable.
+
+For an overview of how Pinokio works and what AI apps can be installed, please refer to the official [Pinokio](https://pinokio.computer) website.
+
+## Building the Container
+To build the container, run:
+```bash
+docker build -t olilanz/ai-pinokio3 .
+```
+
+## Running the Container
+To run the container, use:
+```bash
+docker run -it --rm --name ai-pinokio3 \
+  --shm-size 24g --gpus all \
+  -v /mnt/cache/appdata/ai-pinokio3:/workspace \
+  -p 42000:42000 \
+  --network host \
+  olilanz/ai-pinokio3
+```
+
+## Resources
+* [Pinokio GitHub Issues](https://github.com/pinokiocomputer/pinokio/issues/87)
+* [Pinokio GitHub Issues](https://github.com/pinokiocomputer/pinokio/issues/238)

config.json

@@ -0,0 +1,5 @@
+{
+	"theme": "dark",
+	"home": "/workspace/data",
+	"version": "3.6.0"
+}

startup.sh

@@ -0,0 +1,53 @@
+#!/bin/bash
+
+echo "🔧 Starting Pinokio container startup script..."
+
+# Ensure /run/dbus exists
+mkdir -p /run/dbus
+
+# Start the system D-Bus daemon (needed by Electron)
+echo "🔄 Starting system D-Bus..."
+dbus-daemon --system --fork
+sleep 2  # Give it time to initialize
+echo "✅ System D-Bus started."
+
+# Start the session D-Bus (needed by Electron)
+echo "🔄 Starting session D-Bus..."
+eval $(dbus-launch --sh-syntax)
+export DBUS_SESSION_BUS_ADDRESS
+echo "✅ Session D-Bus started: $DBUS_SESSION_BUS_ADDRESS"
+
+# Start X virtual framebuffer (needed by Pinokio)
+echo "🔄 Starting Xvfb virtual display so Electron has a place to render..."
+Xvfb :99 -screen 0 1280x1024x16 &
+sleep 2
+export DISPLAY=:99
+echo "✅ Xvfb is running (DISPLAY=:99)."
+
+# Setting up the cache directories
+CACHE_HOME="/workspace"
+CONFIG_HOME="${CACHE_HOME}/config"
+PINOKIO_HOME="${CACHE_HOME}/data"
+
+echo "📂 Setting up cache directories..."
+mkdir -p "${CACHE_HOME}" "${CONFIG_HOME}" "${PINOKIO_HOME}"
+
+# Setting up initial config
+if [ ! -f "${CONFIG_HOME}/config.json" ]; then
+    cp "./config.json" "${CONFIG_HOME}/config.json"
+fi
+
+# Redirecting the config to the cache
+mkdir -p "${CONFIG_HOME}" /root/.config 
+ln -snf "${CONFIG_HOME}" /root/.config/Pinokio 
+
+# Start Pinokio (Electron desktop app requires these switches)
+# PINOKIO_ARGS="serve --port 42000 --no-sandbox --disable-gpu --disable-software-rasterizer --disable-dev-shm-usage --headless"
+PINOKIO_ARGS="serve --port 42000 --no-sandbox --disable-dev-shm-usage --headless"
+
+echo "🚀 Starting Pinokio (container port 42000)..."
+echo "exec pinokio ${PINOKIO_ARGS} 2>&1 | tee \"${CACHE_HOME}/output.log\""
+exec pinokio ${PINOKIO_ARGS} 2>&1 | tee "${CACHE_HOME}/output.log"
+
+# If Pinokio exits, print a message
+echo "❌ Pinokio has stopped running. Check logs for details."