Since at least mid-August (maybe earlier), I’ve been frequently running into “Can’t scale up due to the current unavailability of a Compute Engine resource, for example GPUs or CPUs in the requested zone.” This is mostly for n4d-standard-2 machine type, mostly in us-central1 region. I’ve tried adding more zones to my node pools, but I’m still having issues with this machine type. These are relatively small scale-ups of 1 or 2 nodes, but the machines can be unavailable for several hours at a time.
This is creating availability issues for our product, and it seems to me that I really just shouldn’t use this machine family, at least in us-central1. Am I thinking about this wrong? Is Google planning to support this machine type, or should we just stop using it?