# Install dependencies using uv
uv pip sync
# Install Ansible Galaxy roles
uv run ansible-galaxy install --force-with-deps -r requirements.yml

uv run ansible-playbook setup.yml # Update everything

uv run ansible-playbook setup.yml --tags caddy,pihole,homepage --limit homelab,nameserver # update dhcp, domains etc.

uv run ansible-playbook setup.yml --tags ha # Update smart home
uv run ansible-playbook setup.yml --tags ollama # Update ollama

uv run ansible-playbook setup.yml --limit mic-satellites -t satellite-audio
uv run ansible-playbook setup.yml --limit mic-satellites -t wyoming --start-at-task="Start wyoming stack"

uv run ansible-playbook setup.yml --limit virtualhere

ansible-playbook setup.yml --limit ailab
# force ollama container update
ansible-playbook setup.yml --limit ailab -t ollama-arc --extra-vars "force_update=true"
ansible-playbook setup.yml --limit ailab -t ollama-arc --start-at-task "Define ollama container environment variables"

ssh -t ailab.lan 'while true; do sudo docker logs -f --tail=100 ipex-llm-inference-cpp-xpu-container || sleep 10; done'
sudo docker logs -f ipex-llm-inference-cpp-xpu-container
sudo docker exec -it ipex-llm-inference-cpp-xpu-container /bin/bash

ssh ailab.lan
export DOCKER_IMAGE=intelanalytics/ipex-llm-serving-xpu:latest
export CONTAINER_NAME=ipex-llm-serving-xpu-container
sudo docker rm --force $CONTAINER_NAME
sudo docker run -itd \
    --net=host \
    --privileged \
    --device=/dev/dri \
    -v /opt/vllm-models:/llm/models \
    -v ~/.huggingface:/root/.huggingface \
    -e no_proxy=localhost,127.0.0.1 \
    --memory="32G" \
    --name=$CONTAINER_NAME \
    --shm-size="16g" \
    --entrypoint /bin/bash \
    $DOCKER_IMAGE
sudo docker exec -it ipex-llm-serving-xpu-container /bin/bash
cat ./start-vllm-service.sh
sed -i 's#ray#ray \\\n  --download-dir /llm/models \\\n  -q awq \\\n  --kv-cache-dtype fp8 \\\n  --enable-prefix-caching#g' ./start-vllm-service.sh
# sed -i 's/SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=2/SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1/g' ./start-vllm-service.sh
# TODO -q AWQ
# TODO --enable-prefix-caching
# TODO --kv-cache-dtype fp8
vim ./start-vllm-service.sh

export ONEAPI_DEVICE_SELECTOR="level_zero:0"
export NUM_CTX=4096
export IPEX_LLM_NUM_CTX=$NUM_CTX
export MAX_NUM_SEQS=1
export MAX_MODEL_LEN=$NUM_CTX
export MAX_NUM_BATCHED_TOKENS=$((MAX_MODEL_LEN * MAX_NUM_SEQS))
export LOAD_IN_LOW_BIT=woq_int4
export LOAD_IN_LOW_BIT=int4_asym
unset LOAD_IN_LOW_BIT
export TENSOR_PARALLEL_SIZE=1
export PORT=8001

export HF_MODEL=Qwen/Qwen2.5-14B-Instruct
export HF_MODEL=Qwen/Qwen2.5-0.5B-Instruct-AWQ

export HF_MODEL=Qwen/Qwen2.5-0.5B-Instruct
export MODEL_PATH=$HF_MODEL
export SERVED_MODEL_NAME=$HF_MODEL

bash ./start-vllm-service.sh

---

export CCL_WORKER_COUNT=2
export SYCL_CACHE_PERSISTENT=1
export FI_PROVIDER=shm
export CCL_ATL_TRANSPORT=ofi
export CCL_ZE_IPC_EXCHANGE=sockets
export CCL_ATL_SHM=1

export USE_XETLA=OFF
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export ONEAPI_DEVICE_SELECTOR="level_zero:1"
export TORCH_LLM_ALLREDUCE=0

export CCL_SAME_STREAM=1
export CCL_BLOCKING_WAIT=0

sudo docker run -it \
        --rm \
        --network=host \
        --device /dev/dri \
        -v /dev/dri/by-path:/dev/dri/by-path \
        --memory="32G" \
        --shm-size="16g" \
        --entrypoint /bin/bash \
        vllm-xpu:latest

sudo docker run -it \
        --rm \
        --network=host \
        --device /dev/dri \
        -v /dev/dri/by-path:/dev/dri/by-path \
        -v /opt/vllm-models:/models \
        --memory="32G" \
        --shm-size="16g" \
        -e PORT -e CCL_WORKER_COUNT -e FI_PROVIDER -e CCL_ATL_TRANSPORT -e CCL_ZE_IPC_EXCHANGE -e CCL_ATL_SHM -e USE_XETLA -e SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS -e SYNC_CACHE_PERSISTENT -e TORCH_LLM_ALLREDUCE -e CCL_SAME_STREAM -e CCL_BLOCKING_WAIT \
        vllm-xpu:latest \
        --port $PORT \
        --trust-remote-code \
        --enforce-eager \
        --download-dir /models \
        --dtype=float16 \
        --device=xpu \
        --disable-async-output-proc \
        --max-model-len $MAX_MODEL_LEN \
        --max-num-batched-tokens $MAX_NUM_BATCHED_TOKENS \
        --max-num-seqs $MAX_NUM_SEQS \
        --model $HF_MODEL

uv run ansible-vault edit group_vars/all/secrets.yml

ssh pi4
sudo docker exec -it pihole pihole -a -p

ssh nuc
cd /home/nuc
git clone https://github.com/anudeepND/whitelist.git
sudo python3 whitelist/scripts/whitelist.py --dir /home/pi/config/pihole/ --docker
sudo whitelist/scripts/referral.sh --dir /home/pi/config/pihole/ --docker

ssh -A [email protected] sudo rsync -e "ssh" -vuar /media/Medien/{pihole,home-assistant,caddy,mosquitto,vscode,z2m} [email protected]:/home/nuc/config/
ssh [email protected] sudo chown -R nuc:nuc /home/nuc/config
ssh [email protected] .cargo/bin/exa -l --tree --level=2 /home/nuc/config/

uv run ansible-playbook fix_hass_bluetooth.yml

ssh nuc
HOST_PATH=/home/nuc/config/ring-mqtt/
docker run -it --rm --mount type=bind,source=$HOST_PATH,target=/data --entrypoint /app/ring-mqtt/init-ring-mqtt.js tsightler/ring-mqtt
docker restart ring-mqtt

uv run ansible-playbook setup.yml -l mic-satellites -t alsa,pulse
uv run ansible-playbook setup.yml -l mic-satellites -t stack
uv run ansible-playbook setup.yml -l mic-satellites -t vis

ssh micpi pactl set-sink-volume @DEFAULT_SINK@ 80%
ssh micpi amixer set Master 90%
ssh micpi dmesg


ssh micpi 'TERM=xterm vis'
ssh micpi "aplay -D plughw:0,0 /usr/share/sounds/alsa/Front_Center.wav"
ssh micpi "paplay /usr/share/sounds/alsa/Front_Center.wav"
ssh micpi sudo docker exec wyoming-snd-external-alexa aplay /usr/share/sounds/alsa/Front_Center.wav

ssh micpi "sudo systemctl status pulseaudio"
ssh micpi "sudo systemctl restart pulseaudio"
ssh micpi "sudo systemctl stop pulseaudio"
ssh micpi "journalctl -xeu pulseaudio.service"
ssh micpi "pactl info"
ssh micpi "pactl list short sinks"
ssh micpi "pactl list short sources"

osascript <<EOF
tell application "iTerm2"
     create window with default profile
     tell current session of current window
          delay 1
          write text "zellij"
      end tell
end tell

zellij run -- ssh micpi sudo docker logs -f wyoming-snd-external-alexa
zellij run -- ssh micpi sudo docker logs -f wyoming-mic-external-alexa
zellij run -- ssh [email protected] sudo docker logs -f wyoming-satellite-alexa
zellij run -- ssh [email protected] sudo docker logs -f wyoming-porcupine1
zellij run -- ssh micpi tail -f /var/log/jabra_connected.log

ssh micpi "sudo wget -O /boot/firmware/config.txt https://raw.githubusercontent.com/RPi-Distro/pi-gen/refs/heads/master/stage1/00-boot-files/files/config.txt"

ssh micpi "sudo rm -rf /etc/pulse && sudo apt -y purge pulseaudio pulseaudio-utils && sudo apt -y autoremove"