Conversation
Signed-off-by: Mingyu Li <[email protected]>
Signed-off-by: Mingyu Li <[email protected]>
Signed-off-by: Mingyu Li <[email protected]>
…manually run notebooks Signed-off-by: Mingyu Li <[email protected]>
| # Parameters: | ||
| # | ||
| # * env (optional): Name of the environment the notebook is run in (dev, staging, or prod). Defaults to "dev". | ||
| # * env (optional): Name of the environment the notebook is run in (test, staging, or prod). |
There was a problem hiding this comment.
Why not leave dev here? It will be used when the notebook is triggered manually, right?
There was a problem hiding this comment.
We are getting the env field without defining the field in the widget, which means only terraform jobs or github actions will be able to set the env parameter.
If users directly run the notebook, they will directly use the dev profile without being able to set the env field.
vladimirk-db
left a comment
There was a problem hiding this comment.
Can you please test the train/deploy CUJ to make sure nothing is broken?
Also, how will this affect existing users? E.g., we are changing experiment and model names - I assume existing setups won't be broken, can you confirm please?
|
@vladimirk-db Thanks. The testing in CUJ is still in progress. I went through several issues while setting up the monorepo.
I've got the terraform deployed to staging and prod. https://github.com/databricks/mlops-azure-cuj/pull/79 |
Followed option2 in doc https://docs.google.com/document/d/1PO49MfICLpDpCXk5jFq_tcuPbocutotkui30jxSXLXQ/edit#
Tested by adding two projects(receipes & fs) to CUJ repo to make it a monorepo.
The staging and prod workspace workflows succeeded in training, validation, deploy.