Open jhamet93 opened 4 months ago
Hi @jhamet93, thanks for the issue!
How are/would you be doing this in pure Python without Prefect? You can surely form the url of your current job by retrieving your current GCP project & job, as a task, for example, or even set up a state change hook or something if you only care about failing flows.
@serinamarie
This is surely possible within the context of a specific job. For this specific scenario, use your favorite HTTP compatible library to gather the data and compute a valid URL.
I was thinking more along the lines if it would feasible outside of the specific job context and move the responsibility to the business logic that submits the job for execution. This can help debug shorten the time to debug any infrastructure related problems (e.g. why has my job not started yet, my job failed but Prefect hasn't reflected this, etc).
First check
Prefect Version
2.x
Describe the current behavior
When a Flow starts running, it's challenging if need be, to discover the underlying infastructure the Flow is running on if needed be. As a concrete example, I may have a Flow running on Vertex AI that's failing because of memory issues. However, I would need to manually navigate to GCP and find the appropriate job that corresponds to the Flow run.
Describe the proposed behavior
At the beginning of a Flow run, if implemented by different Infrastructure types of a Deployment, log the URL of the infrastructure. For example, Vertex AI implementations like above could hit the metadata server to aggregate the required information to form a fully qualified URL.
Example Use
No response
Additional context
No response