llm-ingestor

This repository contains the following components:

1. Helm Chart for llm-ingestor: Located in /charts/llm-ingestor.
2. Application Code for Ingestor: Located in /ingestor with its Dockerfile.
3. Application Code for LLM Pipeline: Located in /pipelines with its Dockerfile (used by a chart not included in this repository).

Helm Chart Overview

The Helm chart is designed to deploy the llm-ingestor application. Below are the configuration options available in the chart.

Configuration

The following table lists the configurable parameters of the llm-ingestor Helm chart and their default values.

Parameter	Description	Default Value
nameOverride	Name override for the deployment	llm-ingestor
secrets.dockerhubconfigjson	Docker Hub credentials secret	''
env.KAFKA_SERVICE	Kafka service endpoint	localhost:9092
env.KAFKA_TOPIC	Kafka topic to consume	cve
env.KAFKA_CONSUMER_GROUP_ID	Kafka consumer group ID	llm-ingestor
env.KAFKA_MAX_SIZE	Maximum Kafka message size	10000000
env.PORT	Port on which the application listens	3001
env.LLAMAINDEX_OLLAMA_BASE_URL	Base URL for the LlamaIndex Ollama	http://localhost:11434
env.LLAMAINDEX_MODEL_NAME	Model name for LlamaIndex	gemma:2b
env.EMBEDDING_DIM	Embedding dimension for LlamaIndex	1024
env.LLAMAINDEX_EMBEDDING_MODEL_NAME	Embedding model name for LlamaIndex	mxbai-embed-large:latest
initContainers[].name	Name of the init container	db-migrate
initContainers[].repository	Repository for the init container	marlapativ/db-migration
initContainers[].tag	Tag for the init container	latest
initContainers[].pullPolicy	Pull policy for the init container	Always
initContainers[].command	Command to run in the init container	['flyway']
initContainers[].args	Arguments for the init container command	['migrate']
initContainers[].resources.limits.cpu	CPU limit for the init container	100m
initContainers[].resources.limits.memory	Memory limit for the init container	100Mi
initContainers[].resources.requests.cpu	CPU request for the init container	10m
initContainers[].resources.requests.memory	Memory request for the init container	25Mi
extraSecrets	Additional secrets to be provided	{}
pod.readinessProbe.path	Path for the readiness probe	/healthz
pod.livenessProbe.path	Path for the liveness probe	/healthz
resources.limits.cpu	CPU limit for the application pod	500m
resources.limits.memory	Memory limit for the application pod	384Mi
resources.requests.cpu	CPU request for the application pod	250m
resources.requests.memory	Memory request for the application pod	256Mi
podLabels	Labels to be applied to the pod	{}
podAnnotations	Annotations to be applied to the pod	{}
serviceAccount.create	Whether to create a new service account	true
serviceAccount.name	Name of the service account	llm-ingestor-sa
serviceAccount.automount	Automount service account token	false
serviceAccount.annotations	Annotations for the service account	{}
db.secrets.postgresUsername	Secret for PostgreSQL username	''
db.secrets.postgresPassword	Secret for PostgreSQL password	''
db.config.host	Host for the PostgreSQL database	''
db.config.port	Port for the PostgreSQL database	5432
db.config.table	Table name in the PostgreSQL database	embeddings
db.config.schema	Schema name in the PostgreSQL database	cve
kafka.secrets.username	Kafka username	''
kafka.secrets.password	Kafka password	''
replicaCount	Number of replicas for the deployment	3
image.repository	Docker image repository for the application	marlapativ/llm-ingestor
image.tag	Docker image tag for the application	latest
image.pullPolicy	Image pull policy	Always
affinity.enable	Enable affinity for pod scheduling	true
affinity.topologyKey	Topology key for pod affinity	topology.kubernetes.io/zone
pdb.create	Whether to create a Pod Disruption Budget	true
pdb.minAvailable	Minimum number of pods available under PDB	1
pdb.maxUnavailable	Maximum number of pods unavailable under PDB	{}
autoscaling.hpa.enabled	Enable Horizontal Pod Autoscaler	true
autoscaling.hpa.minReplicas	Minimum number of replicas under HPA	1
autoscaling.hpa.maxReplicas	Maximum number of replicas under HPA	3
autoscaling.hpa.targetCPU	Target CPU utilization percentage for HPA	5
autoscaling.hpa.targetMemory	Target memory utilization percentage for HPA	80
autoscaling.hpa.annotations	Annotations for the HPA	{}

Usage

To deploy the llm-ingestor Helm chart, use the following command:

helm install llm-ingestor /charts/llm-ingestor --values /path/to/your/values.yaml

---

This template provides a clear overview of the configuration parameters and their default values, making it easy for users to understand and customize the deployment.