Trouble handling huge data about oryx HOT 3 CLOSED

Clarification: after waiting 10 minutes more I do get responce from oryx, but its irrelevant:

From input data we can see that it is supposed to include i1 and i2, but it doesn't. The server is still calculating something. How long should it take for 100K entries in the database?

from oryx.

One more update, finally computation on server seems to have finished (no CPU usage), but recommendations are still irrelevant:

from oryx.

I don't see the problem in the ingest. The HTTP status is 204, which is correct.

You wouldn't expect to see a new user in that data return recommendations immediately; it would appear after the speed layer had run and pushed updates to the model. There are also corner cases where no recommendations can be made (i.e. items are unknown too).

It looks like the update jobs are still running or something. I'm not sure how much data you have or what you're running on, but if you're trying to put all of this in one small VM it might take a long time. You need to look at the Spark UI to answer those questions.

If a user has interacted with items, they are excluded from recommendations. Right? that's expected.

I can't say whether the recommendations are relevant or not. They are according to the model. You may need to tune the model. For example if rank is too low it will underfit.

from oryx.