startup-script.sh

#!/bin/bash
set -e

echo "Setup started." >/var/log/startup-script.log

DIFY_VERSION="${dify_version}"

# =============================================================================
# Configure Dify
# =============================================================================

# Create .env file from .env.example
cd /opt/dify-$DIFY_VERSION/docker
cp .env.example .env

# Replace configuration values using sed
# Database Configuration
sed -i "s|^DB_HOST=.*|DB_HOST=${db_host}|" .env
sed -i "s|^DB_USERNAME=.*|DB_USERNAME=${database_user}|" .env
sed -i "s|^DB_PASSWORD=.*|DB_PASSWORD='${database_password}'|" .env
sed -i "s|^DB_DATABASE=.*|DB_DATABASE=${database_name}|" .env

# pgvector Configuration
sed -i "s|^VECTOR_STORE=.*|VECTOR_STORE=pgvector|" .env
sed -i "s|^PGVECTOR_HOST=.*|PGVECTOR_HOST=${pgvector_private_ip}|" .env
sed -i "s|^PGVECTOR_USER=.*|PGVECTOR_USER=${pgvector_database_user}|" .env
sed -i "s|^PGVECTOR_PGUSER=.*|PGVECTOR_PGUSER=${pgvector_database_user}|" .env
sed -i "s|^PGVECTOR_PASSWORD=.*|PGVECTOR_PASSWORD='${pgvector_database_password}'|" .env
sed -i "s|^PGVECTOR_POSTGRES_PASSWORD=.*|PGVECTOR_POSTGRES_PASSWORD='${pgvector_database_password}'|" .env
sed -i "s|^PGVECTOR_DATABASE=.*|PGVECTOR_DATABASE=${pgvector_database_name}|" .env
sed -i "s|^PGVECTOR_POSTGRES_DB=.*|PGVECTOR_POSTGRES_DB=${pgvector_database_name}|" .env

# Redis Configuration
sed -i "s|^REDIS_HOST=.*|REDIS_HOST=${redis_host}|" .env
sed -i "s|^REDIS_PORT=.*|REDIS_PORT=${redis_port}|" .env
sed -i "s|^REDIS_PASSWORD=.*|REDIS_PASSWORD='${redis_auth_string}'|" .env
sed -i "s|^CELERY_BROKER_URL=.*|CELERY_BROKER_URL='redis://:${redis_auth_string}@${redis_host}:${redis_port}/1'|" .env

# Disable Default DB
# ETL / Unstructured Configuration
if [ "${enable_unstructured_api}" = "true" ]; then
    sed -i "s|^COMPOSE_PROFILES=.*|COMPOSE_PROFILES=unstructured|" .env
    sed -i "s|^ETL_TYPE=.*|ETL_TYPE=Unstructured|" .env
    sed -i "s|^UNSTRUCTURED_API_URL=.*|UNSTRUCTURED_API_URL=http://unstructured:8000/general/v0/general|" .env
else
    sed -i "s|^COMPOSE_PROFILES=.*|COMPOSE_PROFILES=|" .env
fi

# Password for admin user initialization.
sed -i "s|^INIT_PASSWORD=.*|INIT_PASSWORD='${initial_password}'|" .env

# Comma-separated list of file extensions blocked from upload for security reasons.
sed -i "s|^UPLOAD_FILE_EXTENSION_BLACKLIST=.*|UPLOAD_FILE_EXTENSION_BLACKLIST=exe,bat,cmd,com,scr,vbs,ps1,msi,dll|" .env

chown -R ubuntu:ubuntu /opt/dify-$DIFY_VERSION
echo "Dify was configured." >>/var/log/startup-script.log

# =============================================================================
# Filestore Setup
# =============================================================================

# Mount Filestore to Dify volumes directory
FILESTORE_IP="${filestore_ip}"
FILESTORE_SHARE="${filestore_share_name}"
DIFY_DIR="/opt/dify-$DIFY_VERSION"
VOLUMES_DIR="$DIFY_DIR/docker/volumes"
TEMP_MOUNT="/mnt/filestore_temp"

# Create temporary mount point and mount Filestore
mkdir -p "$TEMP_MOUNT"
mount -t nfs -o rw,intr "$FILESTORE_IP:/$FILESTORE_SHARE" "$TEMP_MOUNT"

# Check if Filestore is empty (only lost+found exists)
ITEM_COUNT=$(ls -A "$TEMP_MOUNT" | grep -v '^lost+found$' | wc -l)
if [ "$ITEM_COUNT" -eq 0 ]; then
    echo "Filestore is empty (only lost+found exists). Copying initial volumes data..." >>/var/log/startup-script.log

    # Copy contents from source volumes directory to Filestore
    if [ -d "$VOLUMES_DIR" ]; then
        cp -a "$VOLUMES_DIR/." "$TEMP_MOUNT/"
        echo "Initial volumes data copied to Filestore." >>/var/log/startup-script.log
    else
        echo "Warning: Source volumes directory not found." >>/var/log/startup-script.log
    fi

    # Additional python packages to be installed in sandbox container.
    mkdir -p "$TEMP_MOUNT/sandbox/dependencies"
else
    echo "Filestore already contains data. Skipping initial copy." >>/var/log/startup-script.log
fi

# Get python-requirements.txt from metadata and save to sandbox volume
curl -s "http://metadata.google.internal/computeMetadata/v1/instance/attributes/python-requirements" \
    -H "Metadata-Flavor: Google" >"$TEMP_MOUNT/sandbox/dependencies/python-requirements.txt"
echo "Python requirements copied to sandbox volume" >>/var/log/startup-script.log

# Get sandbox config from metadata and save to sandbox volume only if python-requirements.txt is not empty
if [ -s "$TEMP_MOUNT/sandbox/dependencies/python-requirements.txt" ]; then
    curl -s "http://metadata.google.internal/computeMetadata/v1/instance/attributes/sandbox-config" \
        -H "Metadata-Flavor: Google" >"$TEMP_MOUNT/sandbox/conf/config.yaml"
    echo "Sandbox config copied to sandbox volume" >>/var/log/startup-script.log
else
    echo "Python requirements is empty. Skipping sandbox config." >>/var/log/startup-script.log
fi

# Unmount temporary mount
umount "$TEMP_MOUNT"
rmdir "$TEMP_MOUNT"

# Remove existing volumes directory if it exists
if [ -d "$VOLUMES_DIR" ]; then
    rm -rf "$VOLUMES_DIR"
fi

# Create mount point and mount Filestore
mkdir -p "$VOLUMES_DIR"
mount -t nfs -o rw,intr "$FILESTORE_IP:/$FILESTORE_SHARE" "$VOLUMES_DIR"

# Add to /etc/fstab for automatic mounting on reboot
echo "$FILESTORE_IP:/$FILESTORE_SHARE $VOLUMES_DIR nfs rw,intr 0 0" >>/etc/fstab

echo "Filestore mounted successfully." >>/var/log/startup-script.log

# =============================================================================
# Start Dify
# =============================================================================

# Start Dify with Docker Compose
sudo -u ubuntu docker compose up -d

echo "Setup completed successfully!" >>/var/log/startup-script.log