matlab AWS cluster exceed vCPU limit despite within quota

Question

I'm trying to start a cloud cluster with AWS on Matlab. I realized I was capped at 32 vcpu so I requested a quota increase to 256 on all normal instances, which was approved.
On matlab cloud center, I asked for 16 workers/machine with machine type m5.8xlarge, 16 core, and requested 16 machines with no dedicated head node, which should put the total vcpu count at 256. However, when I try to start uo the cluster, cloud center keeps telling me that I'm exceeding the maximum vcpu count of 32. Right now I am only able to have 1 machine running. How can I fix this?

Answer

Hello.

Did you apply for a quota in the region where you will launch EC2?  
EC2's vCPU quota is tied to the region, so when applying for quota upper limit relaxation, you need to change it to the region you are applying for.  
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-resource-limits.html  
https://repost.aws/knowledge-center/ec2-instance-limit

You should also check that you applied for the correct quota.  
https://docs.aws.amazon.com/ec2/latest/instancetypes/ec2-instance-quotas.html

For on-demand "m5.8xlarge", you need to apply for the quota shown in the red frame in the image below.  
![a](/media/postImages/original/IMbY_lyWZGSgqfv-pAnD6etw)

matlab AWS cluster exceed vCPU limit despite within quota

Relevant content