I have a question about scheduled jobs. I want to schedule job, that will request backend, fetch ID's that needs to be executed. what's the best option for that?
I came up with such options:
• Nested jobs. One job to have an op that will query all the data I need for a scheduled run
• Two schedules. One will create an asset with ID's, another will pick that meta.
• some super custom partition, that will contain ID and run date
• use GraphQL API to spin up the jobs
Is there a more straight forward option?
the use case is simple, I want to train NLP models for every ID that is enabled a.k.a stored in database every week.
02/08/2023, 12:29 PM
Hi Maksym, I’m not sure I fully understand your scenario, but one possibility is to have one job, one schedule for that job, and then to issue multiple `RunRequest`s from the schedule evaluation function (one for each ID).