[backend] failing to load busybox

[miramar-labs]

Environment

• How did you deploy Kubeflow Pipelines (KFP)?
  minikube, manifests from v1.9.1-rc.3
• KFP version:
  whatever was installed via manifests from v1.9.1-rc.3
  To find the version number, See version number shows on bottom of KFP UI left sidenav. -->
• KFP SDK version:
  kfp 1.1.2
  kfp-pipeline-spec 0.1.16
  kfp-server-api 1.8.5

Steps to reproduce

run a pipline more than once (eg examples/titanic-kaggle)

Expected result

second run should succeed.... without busybox imgPull errors....

Materials and Reference

I'm running kubeflow v1.9.1-rc.3 on minikube and the 'titanic-kaggle' pipeline example.
After one 'mostly' sucessfull run, I corrected an error in a docker image name and tried to re-run the pipeline.
This time, on the very first step (which had previously succeeded), I'm seeing the error:

Failed to pull image "gcr.io/google-containers/busybox": [DEPRECATION NOTICE] Docker Image Format v1 and Docker Image manifest version 2, schema 1 support is disabled by default and will be removed in an upcoming release. Suggest the author of gcr.io/google-containers/busybox:latest to upgrade the image to the OCI Format or Docker Image manifest v2, schema 2. More information at https://docs.docker.com/go/deprecated-image-specs/

Impacted by this bug? Give it a 👍.

[juliusvonkohout]

@rimolive probably you should also upgrade busybox for CVE reasons.

[rimolive]

@miramar-labs The examples repo is very outdated and does not receive much updates recently. I believe this specific example came from an external contributor and used KFP v1 in the example.

Some examples we create is stored in https://github.com/kubeflow/pipelines/tree/master/samples/v2. There is also https://github.com/kubeflow/pipelines/tree/master/samples/contrib but it falls on the same assumptions than the examples repo, they are all created by external contributors and thus does not receive much frequent updates.