Documentation on how ressource are allocated?

[fBedecarrats]

I would like to better understand how SSPCloud handles the min-max resources required for CPUs and Memory. For now I have two simplistic hypotheses on how to interpret it but it might be something else

• Hypothesis 1: a fixed amount of resources is allocated pod startup: it is the maximum if all is available, or less if the maximum is not available, but never less than the set minimum.
• Hypothesis 2: the resources are dynamically adjusted during pod run: when my processes need more resources, they get it unless more resources are not available (above the minimum threshold set at startup). I can never get more resources than set as maximum.

Is one of both hypotheses true? Is it something else? I would love to read more about how this works. I am contributing to the development of packages that use parallelization and we plan to use the future R package for it, but I would need to understand better these things work on the machine I'm using. If anyone has already used future with SSP Cloud, I would love to hear about different strategies (for now I get very puzzling results).

[fcomte]

your hypothesis 2 is the good one ;)