Description
The Dataproc hook now has separate methods for submitting and waiting for job completion.
We should improve this further by adding a method to simply poke the status of the job.
Finally implement a DataprocJobSensor so we can use reschedule mode for jobs known to run for a long time.
Use case / motivation
For Long running dataproc jobs that are known to run for a long time (e.g. > 1 hr) we should not block a slot for the entire duration of the job.
Related Issues
N/A
* Other notes *
@varundhussa is currently working on this for a customer with this use case.
I'd like to work on this. Can you please assign this to me @mik-laj
@varundhussa is currently working on this for a customer with this use case.
Thanks @rootcss and @mik-laj
I have completed the sensor and am completing the test cases. I will raise a PR soon.
Most helpful comment
Thanks @rootcss and @mik-laj
I have completed the sensor and am completing the test cases. I will raise a PR soon.