Health Check Commands: Kubeflow Pipelines

Kubeflow pipelines are reusable end-to-end ML workflows built using the Kubeflow Pipelines SDK. The Kubeflow pipelines service has the following goals:

• End to end orchestration
• Easy experimentation
• Easy re-use

Kubeflow Pipelines consist of the following components:

1. API Server
2. Persistence Agent
3. Scheduled Workflow Controller
4. CRD Viewer
5. Visualization Server
6. UI

Please refer to the following commands to perform a health check or investigation into the Kubeflow Pipeline.

KFP API Server

1. Check the status of the KFP API Server Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline -o wide

2. Inspect the events on the KFP API Server Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline

3. Check the status of the KFP API Server Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline -o wide

4. View KFP API Server logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline -f

KFP Persistence Agent

1. Check the status of the KFP Persistence Agent Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline-persistenceagent -o wide

2. Inspect the events on the KFP Persistence Agent Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline-persistenceagent

3. Check the status of the KFP Persistence Agent Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline-persistenceagent -o wide

4. View KFP Persistence Agent logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline-persistenceagent -f

KFP Scheduled Workflow Controller

1. Check the status of the KFP Scheduled Workflow Controller Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline-scheduledworkflow -o wide

2. Inspect the events on the KFP Scheduled Workflow Controller Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline-scheduledworkflow

3. Check the status of the KFP Scheduled Workflow Controller Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline-scheduledworkflow -o wide

4. View KFP Scheduled Workflow Controller logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline-scheduledworkflow -f

KFP CRD Viewer

1. Check the status of the KFP CRD Viewer Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline-viewer-crd -o wide

2. Inspect the events on the KFP CRD Viewer Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline-viewer-crd

3. Check the status of the KFP CRD Viewer Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline-viewer-crd -o wide

4. View KFP CRD Viewer logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline-viewer-crd -f

KFP Visualization Server

1. Check the status of the KFP Visualization Server Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline-visualizationserver -o wide

2. Inspect the events on the KFP Visualization Server Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline-visualizationserver

3. Check the status of the KFP Visualization Server Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline-visualizationserver -o wide

4. View KFP Visualization Server logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline-visualizationserver -f

KFP UI

1. Check the status of the KFP UI Deployment:

$ kubectl get deployment -n kubeflow ml-pipeline-ui -o wide

2. Inspect the events on the KFP UI Deployment:

$ kubectl describe deployment -n kubeflow ml-pipeline-ui

3. Check the status of the KFP UI Pod:

$ kubectl get pods -n kubeflow -l app=ml-pipeline-ui -o wide

4. View KFP UI logs:

$ kubectl logs -n kubeflow deploy/ml-pipeline-ui -f

Useful links

1. Kubeflow Pipeline Documentation