What would be the best way to have a job trigger another job? For instance, I want to collect a series of changes (job), then I want to have another job propagate those changes according to the type of changes

Hey @Aaron T is there a reason you could have both “jobs” in the same job? ie have an op(s) collect the changes and then an op(s) propagate those changes all within one job?

@jamie I want to have multiple dynamic outputs downstream. Basically, get all items that need to be changed, then make those changes in batches

ok that makes sense. I would recommend using a sensor then. maybe in the first job you can write all of the changes to a bucket. then a sensor can look for new files in the bucket and kick off a job to process the changes

That's an interesting idea. I will take a look at it. Thanks!

by nature, sensors are polling - they run every 30 seconds (configurable) and check for changes vs being told about changes. i don’t think there’s anything wrong with using the graphql client to launch a job though if that works with your architecture/setup better

@jamie what if instead of a job I make an asset, could I use and asset observation to trigger a job?

i don’t believe asset observations can trigger jobs, they’re just to record metadata about an existing asset