Hi folks, I am having issue with CPU quotas. I am using GKE with node pools with autoscaling enabled with preemtible nodes for our DEV… and quota page is showing much higher numbers (for “current” and also for “7day peak”).
Currently I am using 2VMs with total of 8CPUs but quota is showing that all of my 32 CPUs are used and GKE cannot scale up! Any advice or same experience? Thx Petr
1 Like
Hello PetrHanak,
From the looks of it, this could be because you might be using standard CPUs and preemptible CPUs on GKE which if this is true would show up on the quota page as standard CPUS + preemptible CPUs which would potentially make sense to why you would see 32 CPUs being used. For this reason, could you confirm if this is true or not so that we can see if this could potentially be the issue. As if this is not correct, I can try to further investigate the issue to find a root cause and solution.
Best Regards,
Kevin
Google Cloud Platform Support, Montreal
1 Like
Hello Kevin, thanks for your answer. I am looking now what I see in quotas and what we are using.
In reality accross all regions - 10VMs (3 non-preemtible + 7 preemtible) 40 T2D CPUs / 0 other CPUs
In Quotas page - 10VMs / 40 T2D CPUs / 40CPUs (all regions)
So today it seems that T2D CPUs are part of “CPUs” and count is ok
Is somewhere some documentation how the CPU quota works in details? I mean sth. like “time windows”, if T2D is a part of “All CPUs” etc. Thanks a lot
BTW My primary issue was gone over the weekend. 
Hello PetrHanak,
From my understanding, here’s the public documentation I’ve been able to find relating to how CPU quota works in detail (Note however I’m not sure if this is the exact thing your looking for but through my research this is what I’ve been able to find).
Here’s are the public documentations:
1: GKE Quotas: https://cloud.google.com/kubernetes-engine/quotas
2: Compute Engine Quotas (GKE uses Compute Engine Instances so this might be related to your question):