Troubleshoot Amazon AWS EKS deployment issues
Learn how you can troubleshoot various issues when you deploy MOSTLY AI to an AWS EKS cluster. Each of the listed issues includes a description of the problem and solution that shows how to overcome or workaround the issue.
AWS reports VcpuLimitExceeded
Problem
When the deployment script creates a new EC2 instance, you might see the error below.
Status Reason: Could not launch On-Demand Instances. VcpuLimitExceeded - You have
requested more vCPU capacity than your current vCPU limit of 32 allows for the
instance bucket that the specified instance type belongs to.
Please visit http://aws.amazon.com/contact-us/ec2-request to request an adjustment to
this limit. Launching EC2 instance failed.
Solution
You can workaround this issue in one of two ways.
- Request
VcpuLimitExceeded
via http://aws.amazon.com/contact-us/ec2-request (opens in a new tab) - Use smaller EC2 type instances. For example,
m5.xlarge
. Editeks-cluster.yaml
and define the instance type in theWorkerNodesInstanceType
section.WorkerNodesInstanceType: Type: String Description: EC2 instance type for the worker nodes Default: m5.xlarge
Region-specific EKS image
Problem
By default, the deployment script uses the Standard EKS-optimised AMIs in eu-central-1
AWS region. This is defined in the EksAmiIds
section of the eks-cluster.yaml
file
EksAmiIds:
# IDs of the "EKS-optimised AMIs" for the worker nodes:
# https://docs.aws.amazon.com/eks/latest/userguide/retrieve-ami-id.html
# IMPORTANT NOTE: Choose AWS EKS compatible ami IDs only for the region where you are creating the cluster and it's verion (1.27) in our case.
# When ami ID found, update below with region and ami ID.
eu-central-1:
Standard: ami-02ad6c39e1e54ebc8
Solution
If you are deploying the MOSTLY AI App in another AWS region different from eu-central-1
, then you need to find the region-specific AWS AMI ID here: https://docs.aws.amazon.com/eks/latest/userguide/retrieve-ami-id.html (opens in a new tab). To reflect this, add a new region and a standard AMI ID as shown in the line above in eks-cluster.yaml
.