Prerequisites

Hardware

  • 128 GB RAM

  • 16 CPUs

  • 500GB SSD disk

  • 4 V100 or later generation GPUs

Operating System:

  • Ubuntu 18

  • Ubuntu 20

  • DGX OS 5.

Software:

  • Helm 3.0

  • Kubernetes 1.19+

  • Running Kubernetes cluster (example set up guide).

  • Ensure you have a functioning NVIDIA Docker installation. The following command should show the available GPUs:

    sudo docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi

    Response:

Layar Installation Instructions

Log In to Kubernetes Cluster

  1. SSH into your kubernetes cluster with root permissions.

    ssh -i ~/.ssh/private.txt vyasa@12.345.67.89

  2. Login as root.

    sudo su -

  3. Increase the operating system max_map_count:

    echo "vm.max_map_count=262144" > /etc/sysctl.d/99-vyasa.conf
    sysctl -p /etc/sysctl.d/99-vyasa.conf

Fetch and Modify the Helm Chart

  1. In the NGC catalog, search the helm charts tab for the Vyasa Product Suite listing.

  2. To download the helm chart, copy and execute the helm fetch command:

    helm fetch https://helm.ngc.nvidia.com/partners/vyasa/charts/layar-0.1.124.tgz

  3. To install, execute the following command. Change MY_APP_URL to the DNS name or IPV4 IP address of the system you are currently ssh’d into. This is the URL you will use in your browser once all of the Layar images are in a Running status.

    helm install layar ./layar-0.1.124.tgz --set APPURL=MY_APP_URL  --set SSL=false

    Response:

    NAME: layar
    LAST DEPLOYED: Thu Apr 29 21:07:12 2021
    NAMESPACE: default
    STATUS: deployed
    REVISION: 1
    TEST SUITE: None

    NOTES:
    Thank you for installing Layar! The application is starting and should be available in a few minutes at the URL http://1.234.567.891

    Please visit help.vyasa.com for documentation, community forums, and other help.

Monitor the Pods Until All Images are Running

  1. Monitor the pods until all Layar images have been installed properly and are in Running status. You can monitor pod status by executing the following command:

    kubectl get pods | grep vyasa

    Response

    vyasa-tf-serving-ner-ingest-6c47866bbd-qhl4q     1/1  Running  0  16mvyasa-connect-api-6f6d678d78-sd4cr            1/1  Running  0  16mvyasa-connect-cache-5f7d8b6867-8zkqq          1/1  Running  0  16mvyasa-connect-db-64fbc5467b-x6ppb             1/1  Running  0  16mvyasa-cortex-7d8ccff5df-mt5b7                 1/1  Running  0  16mvyasa-data-importer-65cc8f5cb8-vs6mh          1/1  Running  0  16mvyasa-deepchem-754d9db7bc-vbgs6               1/1  Running  0  16mvyasa-document-preview-689cc4c48b-nr2m8       1/1  Running  0  16mvyasa-elk-7bdcb877b4-p469x                    1/1  Running  0  16mvyasa-executor-6bf9ddc767-69dmx               1/1  Running  0  16mvyasa-group-word-phrases-85f5b5b79b-tjnfg     1/1  Running  0  16mvyasa-ide-69c469b7fb-9dtfx                    1/1  Running  0  16mvyasa-index-778758cdcf-swlx5                  1/1  Running  0  16mvyasa-index-visualizer-79fbdfd77c-k99gm       1/1  Running  0  16mvyasa-inference-auth-server-6f9fd5b659-nkndh  1/1  Running  0  16mvyasa-logspout-vntq4                          1/1  Running  0  16mvyasa-manifold-7f4cf8c6fc-nr4lb               1/1  Running  0  16mvyasa-ner-6c8cc4fc6d-mf4qq                    1/1  Running  0  16mvyasa-ner-server-86d6867ff5-sjbsf             1/1  Running  0  16mvyasa-object-store-567b94f9c7-cb46h           1/1  Running  0  16mvyasa-passage-ranking-8496f75468-szt8g        1/1  Running  0  16mvyasa-pr-server-659fdd87-hpf7h                1/1  Running  0  16mvyasa-proxy-57779b7857-rq2sn                  1/1  Running  0  16mvyasa-qa-5b4ff96877-hfjq9                     1/1  Running  0  16mvyasa-qa-server-75b88d5bd4-t4gd9              1/1  Running  0  16mvyasa-query-understanding-7695574fd5-phw5m    1/1  Running  0  16mvyasa-queue-657b7cc99-2mxxl                   1/1  Running  0  16mvyasa-rdkit-6f964b586b-j2vgn                  1/1  Running  0  16mvyasa-rendr-55d5444bbf-vpz9p                  1/1  Running  0  16mvyasa-spark-jobproxy-74c658bc8d-sx6h7         1/1  Running  0  16mvyasa-spark-master-7644d6748b-cn75j           1/1  Running  0  16mvyasa-spark-worker-7bc6d59c6b-qj4jp           1/1  Running  0  16mvyasa-tf-serving-ner-5b4f6fd67b-fsztm         1/1  Running  0  16mvyasa-tf-serving-ner-ingest-6c47866bbd-qhl4q  1/1  Running  0  16mvyasa-triton-server-qa-54f8669556-xmq67       1/1  Running  0  16m 

Launch the Layar Website

  1. Refer back to the installation greeting from the final step in "Fetch and Modify Chart".

  2. Locate the URL in the installation greeting:

  3. Navigate to Layar in Browser. Enter the URL into your preferred browser

    http://1.234.567.891

  4. Register User. Enter the credentials for your first Layar user.

  5. Agree to Terms & Conditions. Click the checkbox to agree to the terms and conditions for the software.

  6. Submit Information. Click the Register button to be taken into the Home page of your Layar data fabric.

Congrats! You are now fully set up with a Layar system and ready to perform deep learning analytics on your data.

To uninstall the Layar helm chart, execute the following command:

helm uninstall layar

Did this answer your question?