Vertex AI Pipeline Run on Persistent Resource

[KazooFanatic]

Hi all,

I'm trying to follow this guide to create a Vertex AI Pipeline run that uses a persistent resource.

However, the `default_runtime` argument specified in the sample is not valid: I get the following error:

 PipelineJob.__init__() got an unexpected keyword argument 'default_runtime'

When I look into the docstring / function documentation I can see that the PipelineJob class indeed doesn't take default_runtime as argument. I have also tried using the Rest endpoint but to no avail.

Has anyone managed to run a VertexAI Pipeline with a persistent resource? Or is there another way I configure this pipeline to use a persistent resource?

Thanks!

[KazooFanatic]

A quick follow-up note to say that apparently the feature is still in beta; so using the Python SDK you'll have to import PipelineJob from aiplatform_v1beta1, or if using the REST API, use the "{region}-aiplatform.googleapis.com/v1beta1" endpoint in order to be able to use.

Documentation is not entirely clear but between the docstring of the PipelineJob class and some educated guesses I was able to get this to work.