0

I am using a Component in a vertex ai pipeline with a custom component.

@component(
    base_image=f"gcr.io..."
)
def temp_step():

It's supposed to have the same library versions during the run of the component in the Vertex AI Pipeline than in the Base Image. But when I try I see that in my docker container I have :

google-api-core==2.11.1
google-auth==2.21.0
google-auth-oauthlib==1.0.0
google-cloud-aiplatform==1.9.0
google-cloud-appengine-logging==0.2.0
google-cloud-audit-log==0.2.5
google-cloud-bigquery==2.26.0
google-cloud-core==2.3.3
google-cloud-logging==2.6.0
google-cloud-storage==1.42.2

VS in the pipeline during the run of the component

google-api-core==2.10.2
 google-api-python-client==1.12.11
 google-auth==1.35.0
 google-auth-httplib2==0.1.0
 google-auth-oauthlib==1.0.0
 google-cloud-aiplatform==1.9.0
 google-cloud-appengine-logging==0.2.0
 google-cloud-audit-log==0.2.5
 google-cloud-bigquery==2.26.0
 google-cloud-core==2.3.3
 google-cloud-logging==2.6.0
 google-cloud-storage==1.42.2

Someone knows why I have this difference ?

I tried to compare both and it was different. I am expecting an explanation why the component is different from the base image ?

Ali Tber
  • 33
  • 4
  • Answering my own question. The kfp library was missing from the requirement so the pipeline compilations was adding it in the component command. And this changed the versions of google packages during the run of the pipeline. Adding kfp in the requirements.txt solved the issue. – Ali Tber Jul 11 '23 at 14:15

1 Answers1

1

Posting this as a community wiki:

As what OP figured out, adding the kfp library in the requirements.txt solved the issue of the versions of the google packages being changed.

Joevanie
  • 489
  • 2
  • 5