Just curious, why is it that the total number of t...
# general
k
Just curious, why is it that the total number of tasks in the logs are always increasing and not identical to the physical plan right from the start?
j
Hmm @Sammy Sidhu do you know?
I thought about this a little more. It has to do with the way that we run our tasks in “batches” at the moment We’re looking at building much better tooling around workload observability. More news on that soon. Hoping to have something similar to the BigQuery UI or Spark UI. Lmk if you’d like to maybe collaborate on that!
k
I actually think having an alternative to the Databricks UI would be cool
j
Can you send me a screenshot of what you’re currently thinking about? Is it the Spark one like this:
k
the logs are within the cell output and split by stages, which you can then click into to see the spark ui
problem though is that the ray jobs now arent in notebooks so probably this would not work unless there's a wrapper that just sends the jobs in and returns the results to the notebook
would be super sleek if all the ray stuff could just happen under the hood though
j
Yeah the end goal for us is to try and abstract Ray from the user as much as possible
We’re thinking something more like the BQ UI:
Easier to understand than the Spark one I think (it mirrors the query plan quite a bit more)
k
Oh yeah I've never seen the bigquery one! Looks nice!