canonical / kfp-operators

Kubeflow Pipelines Operators
Apache License 2.0
2 stars 12 forks source link

`kfp-profile-controller` `custom_images` config default is incorrect for frontend image #484

Closed NohaIhab closed 3 months ago

NohaIhab commented 3 months ago

Bug Description

the default of the custom_images config for kfp-profile-controller charm is:

    default: | 
      visualization_server : ''
      frontend_image : ''

Meanwhile, the actual key used in the json defining the default images is:

{
    "visualization_server": "gcr.io/ml-pipeline/visualization-server:2.0.3",
    "frontend": "gcr.io/ml-pipeline/frontend:2.0.3"
}

This makes the user experience error-prone because if you set in the custom_images config with the key frontend_image rather than frontend, it will not work. This is the case for channels latest/edge and 2.0/stable.

When setting the config as shown in the To Reproduce section below, the charm goes to error, that's not the expected behavior, it should rather log a warning and ignore the config. This happens due to issue #485 in kfp-profile-controller charm.

To Reproduce

juju deploy kfp-profile-controller --trust --channel=2.0/stable

juju config kfp-profile-controller custom_images

#output
  custom_images: 
    default: |
      visualization_server : ''
      frontend_image : ''

# set the image as the default tells you
juju config kfp-profile-controller custom_images="visualization_server: 'gcr.io/ml-pipeline/visualization-server:2.0.3'
frontend_image: 'gcr.io/ml-pipeline/visualization-server:2.0.3'
"

# check juju status: charm goes to error
kfp-profile-controller/0*     error

Environment

1.25-strict/stable juju 3.1/stable

Relevant Log Output

-

Additional Context

No response

syncronize-issues-to-jira[bot] commented 3 months ago

Thank you for reporting us your feedback!

The internal ticket has been created: https://warthogs.atlassian.net/browse/KF-5755.

This message was autogenerated