We are interested in splitting up our content cooks by map across multiple build agents. If we do that, is there a way to combine the cook results from multiple build agents before performing staging and packaging?
We are aware that multi-process cook does not currently support remote workers yet. We would be doing the orchestration of the distributed cooks in our build environment ourselves.
There is no way to do this currently. It requires copying the cooked files into a unioned cooked output folder and merging the artifacts that collect data from all packages. The artifacts in engine that do so are currently the shader library and the assetregistry.
We have implemented this merge locally for Fortnite, but have not yet pushed that into engine; no ETA yet for releasing a robust version of it. If you’re interested in doing the work yourself before we make a robust engine version of it, I can give you the list of the problems we encountered and how we solved them. In that case I suggest you take 5.6 before starting the work, and 5.7 as soon as it comes out (ETA Nov 12, preview is available now), to avoid the need for changes in the engine.