CentOS 7 EE GPU Install With Tarball

Note MapD has been rebranded to OmniSci.

This is an end-to-end recipe for installing OmniSci Enterprise Edition on a CentOS 7 machine running with NVIDIA Volta, Kepler, or Pascal series GPU cards using a tarball.

Here is a quick video overview of the installation process.

The installation phases are:
Important The order of these instructions is significant. To avoid problems, install each component in the order presented.

Assumptions

These instructions assume the following:
  • You are installing on a “clean” CentOS 7 host machine with only the operating system installed.
  • Your OmniSci host only runs the daemons and services required to support OmniSci.
  • Your OmniSci host is connected to the Internet.

Preparation

Prepare your Centos 7 machine by installing JDK and EPEL, updating your system, creating the OmniSci user (named mapd), installing CUDA, and enabling a firewall.

JDK

Follow these instructions to install a headless JDK and configure an environment variable with a path to the library. The “headless” Java Development Kit does not provide support for keyboard, mouse, or display systems. It has fewer dependencies and is best suited for a server host. For more information, see https://openjdk.java.net.

  1. Open a terminal on the host machine.
  2. Install the headless JDK using the following command:
    sudo yum install java-1.8.0-openjdk-headless

EPEL

Install the Extra Packages for Enterprise Linux (EPEL) repository. RHEL-based distributions require Dynamic Kernel Module Support (DKMS) to build the GPU driver kernel modules. For more information, see https://fedoraproject.org/wiki/EPEL.

If necessary, download the EPEL package for your RHEL-based CentOS version.

  • RHEL/CentOS 7 64-Bit
    wget http://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm
    rpm -ivh epel-release-latest-7.noarch.rpm
  • RHEL/CentOS 6 32-Bit
    wget http://dl.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm
    rpm -ivh epel-release-6-8.noarch.rpm

If wget is not installed in your environment, use sudo yum install wget to install it.

Use Yum to install the epel-release package.

sudo yum install epel-release

Update and Reboot

Update the entire system and reboot to activate the latest kernel.

sudo yum update
sudo reboot

Create the OmniSci User

Create a group called mapd and a user named mapd, who will be the owner of the OmniSci database. You can create both the group and user with the useradd command and the -U switch.

sudo useradd -U mapd

Install CUDA Drivers

CUDA is a parallel computing platform and application programming interface (API) model. It uses a CUDA-enabled graphics processing unit (GPU) for general purpose processing. The CUDA platform provides direct access to the GPU virtual instruction set and parallel computation elements. For more information on CUDA unrelated to installing OmniSci, see http://www.nvidia.com/object/cuda_home_new.html.

Prepare to Install CUDA Drivers

  1. Identify the Linux kernel you are using by issuing the uname -r command.
  2. Use the name of the kernel (3.10.0-862.11.6.el7.x86_64 in the following code example) to install kernel headers and development packages:
    sudo yum install kernel-devel-3.10.0-862.11.6.el7.x86_64 kernel-headers-3.10.0-862.11.6.el7.x86_64

Install the Drivers

OmniSci requires only the CUDA drivers and not the entire CUDA package. To install the drivers:

  1. Go to https://developer.nvidia.com/cuda-downloads.
  2. Select the target platform by selecting the operating system (Linux), architecture, distribution, version, and installer type (OmniSci recommends rpm (network)).

    CUDA install

  3. In Download Installer..., right-click the Download button and copy the link location of the Base Installer. Do not use the installation instructions on the CUDA site:

    CUDA base installer

  4. Use one of the following methods to download the installer from the command line, using the download link you copied (https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-repo-rhel7-10.0.130-1.x86_64.rpm, in this example):
    • curl:
      curl -O https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-repo-rhel7-10.0.130-1.x86_64.rpm
    • wget:
      wget https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-repo-rhel7-10.0.130-1.x86_64.rpm
  5. Install the CUDA drivers:
    sudo rpm --install cuda-repo-rhel7-10.0.130-1.x86_64.rpm
    sudo yum clean expire-cache
    sudo yum install cuda-drivers
  6. Reboot your system to ensure that all changes are active.
    sudo reboot
    
Note You might see the following warning:
warning: cuda-repo-rhel7-10.0.130-1.x86_64.rpm: Header V3 RSA/SHA512 Signature, key ID 7fa2af80: NOKEY
Ignore it for now; you can verify CUDA driver installation at the Checkpoint.

Checkpoint

  1. Go to /usr/lib64/ and verify that the file libcuda.so is in that location.
  2. Run nvidia-smi to verify that your drivers are installed correctly and recognize the GPUs in your environment. Depending on your environment, you should see something like this to verify that your NVIDIA GPUs and drivers are present:

    NVIDIA SMI

  3. If you see an error like the following, the NVIDIA drivers are probably installed incorrectly:
    NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running
    Review the Install CUDA Drivers section and correct any errors.

Firewall

To use Immerse, you must prepare your host machine to accept HTTP connections. You can configure your firewall for external access.

sudo firewall-cmd --zone=public --add-port=9092/tcp --permanent
sudo firewall-cmd --reload

For more information, see https://fedoraproject.org/wiki/Firewalld?rd=FirewallD.

Installation

You install the OmniSci application itself by expanding a TAR file.

  1. Download the desired version of OmniSci from the URL provided you by your Account Executive.
  2. Create your $MAPD_PATH directory. OmniSci recommends /opt/mapd.
  3. Expand the OmniSci archive file in the $MAPD_PATH directory with the following command:
    tar -xvf <file_name>.tar.gz

Configuration

These are the steps to prepare your OmniSci environment.

Set Environment Variables

For convenience, you can update .bashrc with the required environment variables.

  1. Open a terminal window.
  2. Enter cd ~/ to go to your home directory.
  3. Open .bashrc in a text editor. For example, sudo gedit .bashrc.
  4. Edit the .bashrc file. Add the following export commands under “User specific aliases and functions.”

    # User specific aliases and functions
    export MAPD_USER=mapd
    export MAPD_GROUP=mapd
    export MAPD_STORAGE=/var/lib/mapd
    export MAPD_PATH=/opt/mapd
    export MAPD_LOG=/var/lib/mapd/data/mapd_log
  5. Save the .bashrc file.
  6. Open a new terminal window to use your changes.

The $MAPD_STORAGE directory must be dedicated to OmniSci: do not set it to a directory shared by other packages.

OmniSci Configuration File (mapd.conf)

You can also create a configuration file with optional settings. See Configuration.

Initialization

These are the steps to initialize the database and prepare systemd commands for OmniSci.

This step initializes the database and prepares systemd commands for OmniSci.

  1. Run the systemd installer. This script requires sudo access. You might be prompted for a password. Accept the values provided (based on your environment variables) or make changes as needed. The script creates a data directory in $MAPD_STORAGE with the directories mapd_catalogs, mapd_data, and mapd_export. mapd_import and mapd_log directories are created when you insert data the first time. The mapd_log directory is the one of most interest to a OmniSci administrator.

    cd $MAPD_PATH/systemd
    ./install_mapd_systemd.sh
    

Activation

Start and use OmniSci Core and Immerse.

  1. Start OmniSci Core

    sudo systemctl start mapd_server
    sudo systemctl start mapd_web_server
    
  2. Enable OmniSci Core to start when the system reboots.

    sudo systemctl enable mapd_server
    sudo systemctl enable mapd_web_server
    

Checkpoint

To verify that all systems are go, load some sample data, perform a mapdql query, and generate a pointmap using Immerse.

OmniSci ships with two sample datasets of airline flight information collected in 2008. To install the sample data, run the following command.

cd $MAPD_PATH
./insert_sample_data

When prompted, choose whether to insert dataset 1 (7 million rows) or dataset 2 (10 thousand rows).

Enter dataset number to download, or 'q' to quit:
#     Dataset           Rows    Table Name          File Name
1)    Flights (2008)    7M      flights_2008_7M     flights_2008_7M.tar.gz
2)    Flights (2008)    10k     flights_2008_10k    flights_2008_10k.tar.gz

Connect to OmniSci Core by entering the following command in a terminal on the host machine (default password is HyperInteractive):

$MAPD_PATH/bin/mapdql
password: ••••••••••••••••

Enter a SQL query such as the following:

mapdql> SELECT origin_city AS "Origin", dest_city AS "Destination", AVG(airtime) AS
"Average Airtime" FROM flights_2008_10k WHERE distance < 175 GROUP BY origin_city,
dest_city;
Origin|Destination|Average Airtime
Austin|Houston|33.055556
Norfolk|Baltimore|36.071429
Ft. Myers|Orlando|28.666667
Orlando|Ft. Myers|32.583333
Houston|Austin|29.611111
Baltimore|Norfolk|31.714286

Connect to Immerse using a web browser connected to your host machine on port 9092. For example, http://omnisci.mycompany.com:9092.

Create a new dashboard and a Scatter Plot to verify that backend rendering is working.

  1. Click New Dashboard.
  2. Click Add Chart.
  3. Click SCATTER.
  4. Click Add Data Source.
  5. Choose the flights_2008_10k table as the data source.
  6. Click X Axis +Add Measure.
  7. Choose depdelay.
  8. Click Y Axis +Add Measure.
  9. Choose arrdelay.

The resulting chart shows, unsurprisingly, that there is a correlation between departure delay and arrival delay.

4_firstScatterplot.png