1 Answer
- Newest
- Most votes
- Most comments
0
Ensure your EKS cluster's auto-scaling is set correctly. Verify that your Cluster Autoscaler is properly configured and your pod ResourceRequests and ResourceLimits are set to avoid resource contention. Also, ensure that your EC2 Auto Scaling Groups are configured to span multiple Availability Zones for better resilience and resource utilization. Make sure you're not hitting any Spot Instance capacity limits if using Spot Instances, and diversify your instance types to increase your chances of scaling successfully.
Relevant content
- asked 8 months ago
- asked a year ago
- asked 6 months ago
- AWS OFFICIALUpdated 10 months ago
- AWS OFFICIALUpdated 5 months ago
- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated 2 months ago
Adding on to the last part of this: Go to the ASG and look at its Activity History. Are there launch failures due to spot capacity? Spot best practices are to use at least a dozen different instance types and as many AZs as possible to reduce the chances of capacity related failures. Remember though that CAS in K8s requires all instance types to have the same amount of vCPUs, RAM, etc. So you might not be able to get a full dozen+ instance types, but you could at least add t3a.micro and t2.micro