Deployment

Deploy RAG Pipeline

I. Deploy to Openshift or Kubertenes

Prerequisites

(Change commands to deploy to another cloud)

• A Kubernetes or Red Hat OpenShift cluster on which you can deploy an application
• A MongoDB server
• Reranker service ( we are using DrDecr model from primeQA)
• Retriever ( Elasticsearch or solr or Watson Discovery)
• Python 3.9 or later

1. Build the docker image

Clone the github repo

git clone https://github.com/EnterpriseLLM/SuperKnowa.git
cd 8.\ Application/Backend/

docker build -t <IMGAGE_REGISTRY>/<NAMESPACE>/<IMAGE_NAME>:<IMAGE_TAG>

push the image to the image registry

2. Prepeare configuration

Create a secret

oc create secret generic superknowa-backend-config \
--from-literal MODEL_SERVICE_URL=<Watsonx.ai workbench api url> \
--from-literal MODEL_AUTH_TOKEN='Bearer <Auth token>' \
--from-literal WD_URL=<Watson Discovery URL> \
--from-literal WD_AUTH_KEY=<Watson Discovery authentication key> \
--from-literal WD_PROJECT_ID='<Watson Discovery project id>' \
--from-literal WD_COLLECTION_ID='<Watson Discovery collection id>' \
--from-literal SOLR_BACKEND=<Solr server url> \
--from-literal RERANKER_URL=<Reranker server url> \
--from-literal MONGO_URL='<MONGODB Server url>' \
--from-literal MONGO_PASS='<MongoDB password>' \
--from-literal MONGO_DB=<MongoDB Database name> \
--from-literal ELASTIC_URL='<Elasticsearch url with user and password>'

Create a secret for mongodb certificate and elasticsearch certificate

oc create secret generic superknowa-ca-certificate \
--from-file elastic.crt=cert/elastic.crt \
--from-file mongo.crt=cert/mongo.crt

3. Deploy Applications

cd .../Deployment

oc apply -f deployment.yaml

oc apply -f services.yaml