CrackQ-Ubuntu-setup-notes.txt

# Setup notes from deployment of CrackQ on Google Cloud Ubuntu host with Nvidia T4 GPUs - August 2023

# Install GPU drivers and verify "hardware"

## GPU Driver installation for Nvidia T4
https://cloud.google.com/compute/docs/gpus/install-drivers-gpu#installation_scripts

# Unlikely, but if you already have the Ops Agent collecting GPU metrics on your VM, you must stop the agent before you can install or upgrade your GPU drivers using this installation script.
# After you have completed the installation or upgrade of the GPU driver, you must then reboot the VM.

sudo systemctl stop google-cloud-ops-agent

# Install via script
mkdir setup-stuff
cd setup-stuff/
curl https://raw.githubusercontent.com/GoogleCloudPlatform/compute-gpu-installation/main/linux/install_gpu_driver.py --output install_gpu_driver.py
sudo python3 install_gpu_driver.py
# script might restart your VM (did not for me). If the VM restarts, run the script again to continue the installation.

# verify the GPUs are recognized at the host level
sudo nvidia-smi

# if this command fails, verify that GPUs are attached to the VM (
sudo lspci | grep -i "nvidia"

# if the GPUs are present,  check for kernel/driver mismatch


# install Google's GPU-aware ops agent


#  CrackQ install 
cd  /etc
sudo git clone https://github.com/f0cker/crackq.git
cd crackq


# From https://github.com/f0cker/crackq/wiki/Install-on-Ubuntu
# Install Docker CE

sudo apt-get install \
    apt-transport-https \
    ca-certificates \
    curl \
    software-properties-common

curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
sudo add-apt-repository \
   "deb [arch=amd64] https://download.docker.com/linux/ubuntu \
   $(lsb_release -cs) \
   stable"
sudo apt-get update
sudo apt-get install docker-ce



# Ubuntu Nvidia Runtime Install.   The Nvidia Runtime container toolkit is required to pass the GPUs through to the Docker containers.

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit
sudo systemctl restart docker
sudo apt-get install nvidia-container-runtime

add the following to your /etc/docker/daemon.json file:

{
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}


# Run the installation script.   This will build out some stuff (including the folders under /var/crackq)
sudo ./install.sh docker/nvidia/ubuntu


# Personal preferences for setup.   Feel free to deviate here.

# get the UUID of the partition
sudo blkid

sudo mkdir /var/crackq/files/wordlists
sudo mount /dev/sdb1  /var/crackq/files/wordlists  
# to make permanent:  
sudo vi /etc/fstab 

# example
# [Device]          [Mount Point]                [File System Type] [Options] [Dump] [Pass]
# UUID=xxx.yyy.zzz  /var/crackq/files/wordlists   auto              defaults    0       2

# app setup
# Copy the generic app config into the dest,  generate a new key, and edit.
sudo cp ./crackq.conf /var/crackq/files/
python3 -c 'import secrets; print(secrets.token_urlsafe())'
sudo vi /var/crackq/files/crackq.conf

sudo chown crackq:crackq /var/crackq/files/crackq.conf
sudo chmod 640 /var/crackq/files/crackq.conf


# Copy the generic ngingx
sudo cp ./cfg/crackq_nginx.conf /var/crackq/files/nginx/conf.d    ??????

# create key and CSR
cd /var/crackq/files/nginx/conf.d/
sudo openssl req -newkey rsa:2048 -nodes -keyout <NAME>.key -out <name>.csr 

# ls -l
-rw-r--r--. 1 root   root   5706 Jul 14 14:52 certificate.pem
-rw-r--r--. 1 root   root   1033 Jul 14 14:41 crackq-cert.csr
-rw-r--r--. 1 crackq crackq 2121 Jul 12 21:37 crackq_nginx.conf
-rw-------. 1 root   root   1704 Jul 14 14:40 private.pem

# set the server_name
sudo vi /var/crackq/files/nginx/conf.d./crackq_nginx.conf

# Per NOTICE: Containers losing access to GPUs with error: "Failed to initialize NVML: Unknown Error"     # https://github.com/NVIDIA/nvidia-docker/issues/1730
sudo nvidia-ctk system create-dev-char-symlinks  --create-all

# CD into the crackq install folder  (where you cloned the git repo,  not the /var/crackq)

# start interactively once - verify the page loads in a browser (can't log in yet)
sudo docker compose -f docker-compose.nvidia.yml up 
# start containers as daemon
sudo docker compose -f docker-compose.nvidia.yml up -d


# check that hashcat can access the GPUs
sudo docker exec -it crackq hashcat -I

# if hashcat doesn't see the GPUs,  try editing the crackq container def to use devel flavor of NVIDIA image
# see https://github.com/f0cker/crackq/issues/40

# create a SQL-auth user for UI access
sudo docker exec -it crackq /usr/bin/python3 /opt/crackq/build/crackq/add_admuser.py

# Log in and run a benchmark to verify the GPUs