What happened?

Job groups are as of writing not generally usable in the public interface, but Query on Batch now uses them to structure actions and stages of computation. For example, a Hail Query session will have one Batch with the following job group structure:

1. read_table job group
    - driver read_table job
2. execute job group
    - driver job
    - stage 1 job group
        - partition 1 job
        - partition 2 job
        …
    - stage 2 job group
        - partition 1 job
        - partition 2 job
        …

but the current batch UI just has a jobs table listing all of the jobs in order of submission. It would be a huge boon to UX for users to be get a high level view of the structure of their query and to contextualize a failed job through its action > stage > partition structure.

This should be an almost entirely front end change. It will likely involve additional HTTP endpoints but should not affect the batch data model or the way QoB engages with it.

Version

0.2.131

Relevant log output

No response