Implementation:Kubeflow Pipelines Kubernetes PVC Sample

Knowledge Sources	Kubeflow_Pipelines KFP Kubernetes Platform
Domains	Pipeline_Sample, Kubernetes, Storage
Last Updated	2026-02-13 14:00 GMT

Overview

Sample pipeline demonstrating the full lifecycle of Kubernetes PersistentVolumeClaim usage within a KFP v2 pipeline using the `kfp-kubernetes` plugin.

Description

This sample (62 lines) demonstrates creating, mounting, and deleting a PVC within a pipeline. Two components are defined: `make_data` writes to a volume mount, `read_data` reads from it. The pipeline creates a 5Gi PVC with ReadWriteOnce access via `kubernetes.CreatePVC`, mounts it to both tasks at different paths using `kubernetes.mount_pvc`, enforces ordering via `.after()`, and cleans up with `kubernetes.DeletePVC`.

Usage

Reference this sample when building pipelines that require volume-based data sharing between steps, especially for large datasets where artifact-based passing is impractical.

Code Reference

Source Location

• Repository: Kubeflow_Pipelines
• File: samples/core/kubernetes_pvc/kubernetes_pvc.py
• Lines: 1-62

Signature

@dsl.component
def make_data():
    """Writes data to /data/file.txt on the mounted PVC."""

@dsl.component
def read_data():
    """Reads data from /reused_data/file.txt on the mounted PVC."""

@dsl.pipeline
def my_pipeline():
    """Pipeline: CreatePVC -> make_data -> read_data -> DeletePVC."""

Import

from kfp import dsl
from kfp import kubernetes

I/O Contract

Inputs

Name	Type	Required	Description
PVC size	string	Yes	Storage size (default: "5Gi")
Access mode	string	Yes	PVC access mode (default: "ReadWriteOnce")

Outputs

Name	Type	Description
Compiled YAML	file	Pipeline IR YAML for submission to KFP

Usage Examples

Full Pipeline Code

from kfp import dsl
from kfp import kubernetes

@dsl.component
def make_data():
    with open("/data/file.txt", "w") as f:
        f.write("Hello from make_data")

@dsl.component
def read_data():
    with open("/reused_data/file.txt", "r") as f:
        print(f.read())

@dsl.pipeline
def my_pipeline():
    pvc = kubernetes.CreatePVC(
        pvc_name_suffix="-my-pvc",
        size="5Gi",
        access_modes=["ReadWriteOnce"],
    )
    task1 = make_data()
    kubernetes.mount_pvc(task1, pvc_name=pvc.outputs["name"], mount_path="/data")

    task2 = read_data().after(task1)
    kubernetes.mount_pvc(task2, pvc_name=pvc.outputs["name"], mount_path="/reused_data")

    kubernetes.DeletePVC(pvc_name=pvc.outputs["name"]).after(task2)

Related Pages

• Environment:Kubeflow_Pipelines_Python_SDK