Question: how to expose all features?

[vsoch]

Hi! I'm hoping this is a simple question to answer. How do I get NFD to expose all labels in Kubernetes? When I install out of the box, there are somewhere are 70 labels, however when you just export the raw features there are close to 10k. I'd like to get all 10K exported for my cluster. An I assume these are a set that were deemed important (but was this experimentally derived)?

I'm aware that NFD has black/white lists, and I was expecting to find some filtering criteria in the standard config to install. However, I don't see anything obvious, and I'm guessing there is an automatic subset creation somewhere in the code. Could you advise how to get the worker to export all features available? Thanks!

[vsoch]

It appears you can't - if you set the white list to match all, you only get a little over 100. For the export labels/feature PR I am working on, exporting raw features != exporting labels.

$ bin/nfd export labels | wc -l
127

bin/nfd export features| wc -l
11484

I'd ideally want to use the larger set for experiments. But then if/when I find a meaningful set, it's not clear that would be provided as a label.