Corey was looking for a way to add a verify/validation step after the pipeline runs to examine postgres predictions.
This cannot be a task in the pipeline because the predictions, and run are only written toipostgres after the pipeline is completed.
The intent is to provide an optional post-pipeline task/tasks that can verify/validate without the race conditions and complexity invited by:
an additional script that must be called after a run is complete.
an additional microservice that must be triggered or listen for the run to finish.
@mdbecker Let me know if you have other ideas to handle validation systems test: ingest -> predict -> postgres. I don't want to test the whole integration: ingest -> predict -> postgres -> epic. The validation/verification is to check the model (not the entire integration pipeline). Ideally we could exclude ingest errors as well, but not at this point.
Corey was looking for a way to add a verify/validation step after the pipeline runs to examine postgres predictions.
This cannot be a task in the pipeline because the predictions, and run are only written toipostgres after the pipeline is completed.
The intent is to provide an optional post-pipeline task/tasks that can verify/validate without the race conditions and complexity invited by:
@mdbecker Let me know if you have other ideas to handle validation systems test: ingest -> predict -> postgres. I don't want to test the whole integration: ingest -> predict -> postgres -> epic. The validation/verification is to check the model (not the entire integration pipeline). Ideally we could exclude ingest errors as well, but not at this point.