In my never-ending quest to ensure that PDTs are building as expected, I’ve used the /explore/i__looker/pdt_log explore to put together a number of daily reports that me and my team use to ensure things are running smoothly.
I’m fairly certain we could build even more effective tooling if we could use this data in our own view and, subsequently (and awesomely), make PDTs out of this data. And now I’ve searched thoroughly enough to feel it’s time to actually ask here 😃
Is there a way to query Looker’s internal database and with that tables such as pdt_event_log? Is this even stored in “our” database (in our case Redshift) and just tidily tucked away or is it hosted SaaS-style along with our Looker instance?
Your hunch is correct— The i__looker data is not stored on your database, but rather a Looker-specific database that runs alongside the Looker instance.
You can’t currently build your own views and models off of i__looker, though it is feedback we’ve heard before. We definitely have customers that ETL their i__looker data to their own database on a regular schedule, and then do with it what they will. That’d be the easiest way for you to take control of the Looker metadata and build your own views on top of it.
Izzy, what an awesome, no-brainer-ish idea! I imagine I’ll set up a Look from the pdt_log explore and then use a simple Lambda function to transport the data from the Looker API to Redshift.
I specialize in ideas that require very little to no brain 🙂
That sounds like a great workflow, happy ETLing!