search

databricks-demos / dbdemos

Demos to implement your Databricks Lakehouse
Other
281 stars 91 forks source link

lakehouse-hls-readmission should be installed using non-serverless machine learning runtime compute #147

Open KhagayN opened 1 week ago

KhagayN commented 1 week ago
  1. When installing the lakehouse-hls-readmission solution accelerator with serverless compute, many issues with the codebase were encountered. The DLT pipeline failed because it was utilizing serverless compute by default. I instead needed to disable serverless for it to default to non-serverless compute. That got the DLT pipeline working, but the debugging of this was unnecessary, because if I utilized a non-serverless ML runtime cluster to install lakehouse-hls-readmission, the DLT pipeline runs successfully.

There should be clear instructions on what type of cluster to install the solution accelerator.

  1. The masking function cannot be applied to a streaming table in the governance notebook. Screenshot 2024-10-22 at 10 23 47 PM
QuentinAmbard commented 5 days ago

hey, fyi we're doing a full update for serverless so this should be solved in a few weeks. We had a few advanced issues that we're working on with the product, so it got a bit delayed but it's coming soon!