Step 12 - Node Configuration
  • 24 Jan 2024
  • 1 Minute to read
  • Dark
    Light
  • PDF

Step 12 - Node Configuration

  • Dark
    Light
  • PDF

Article summary

Update

24/01/2024: Document initialization (Ops)

Procedure for Configuring Nodes for Paradigm Deployment

Work in progress - Node configuration.jpg

This section explains how to configure nodes for deploying the Paradigm application using Kubernetes.

Configuration for Single Node Deployment

  • If you prefer to deploy all Paradigm components on a single node (LLM, Paradigm, Redis, Celery, PostgreSQL), check the Paradigm Single Node Deployment box.
    • This option simplifies the configuration by applying the same node name to all components.

Configuration for Multi-Node Deployment

  • If you are using multiple nodes, you will need to specify the name of each node for the following components:
    • Node Name for LLM-serving: Enter the Kubernetes node name for the LLM service.
    • Node Name for Redis: Specify the node for Redis.
    • Node Name for Paradigm: Indicate the node where Paradigm will run.
    • Node Name for Celery: Define the node for Celery.
    • Node Name for Postgres: Name the node dedicated to PostgreSQL.

Important Tips and Remarks

  • The node names must match the kubernetes.io/hostname attribute used in Kubernetes.
  • This configuration will determine the pod affinity, i.e., how pods are distributed among nodes.

Saving and Applying Settings

  • After entering the node names, ensure that all information is accurate.
  • Save the settings to set up the node affinity and pod distribution for your Paradigm deployment.
Final Note

Configuring nodes is a crucial step in ensuring that Paradigm components are deployed on the desired infrastructure. Proper planning of pod affinity can improve the performance and reliability of the application.


Was this article helpful?