Implementation:Kubeflow Pipelines Install Config

Knowledge Sources	Kubeflow_Pipelines
Domains	Deployment, Configuration
Last Updated	2026-02-13 14:00 GMT

Overview

Central ConfigMap defining database settings, caching behavior, artifact storage, and deployment-wide parameters for KFP installation.

Description

The pipeline-install-config ConfigMap (109 lines) is the single source of truth for KFP deployment configuration. It contains: MySQL connection info (dbHost, dbPort: 3306), database names (mlmdDb, cacheDb, pipelineDb), artifact bucket name, default pipeline root, cache staleness limits, cache image, cron timezone, connection lifetime, log level, artifact proxy toggle, and artifact retention days. Referenced by multiple components via environment variable injection.

Usage

Applied during KFP deployment. Operators modify this ConfigMap to configure database connections, storage backends, and caching behavior for their specific environment.

Code Reference

Source Location

• Repository: Kubeflow_Pipelines
• File: manifests/kustomize/base/installs/generic/pipeline-install-config.yaml
• Lines: 1-109

Signature

apiVersion: v1
kind: ConfigMap
metadata:
  name: pipeline-install-config
data:
  dbHost: "mysql"
  dbPort: "3306"
  mlmdDb: "metadb"
  cacheDb: "cachedb"
  pipelineDb: "mlpipeline"
  bucketName: "mlpipeline"
  defaultPipelineRoot: ""
  cacheImage: "ghcr.io/containerd/busybox"
  ConMaxLifeTime: "120s"
  ARTIFACT_COPY_STEP_CABUNDLE_CONFIGMAP_NAME: ""
  ARTIFACT_COPY_STEP_CABUNDLE_CONFIGMAP_KEY: ""
  ARTIFACT_COPY_STEP_CABUNDLE_MOUNTPATH: ""

Import

kubectl apply -f pipeline-install-config.yaml

I/O Contract

Inputs

Name	Type	Required	Description
Database credentials	ConfigMap data	Yes	MySQL host, port, database names
Storage config	ConfigMap data	Yes	Bucket name, pipeline root path

Outputs

Name	Type	Description
Environment variables	env injection	Consumed by cache server, API server, metadata writer via ConfigMap refs

Usage Examples

Customizing for Production

apiVersion: v1
kind: ConfigMap
metadata:
  name: pipeline-install-config
data:
  dbHost: "cloudsql-proxy"
  dbPort: "3306"
  mlmdDb: "prod_metadb"
  cacheDb: "prod_cachedb"
  pipelineDb: "prod_mlpipeline"
  bucketName: "my-prod-bucket"
  defaultPipelineRoot: "gs://my-prod-bucket/pipeline-root"

Related Pages

• Environment:Kubeflow_Pipelines_KFP_Backend_Deployment