[EKS] [Feature Request]: EKS Optimized AMI has very low open-files ulimit

[trimbleAdam]

Tell us about your request
Update EKS optimized AMI ulimits nofile should be far greater than what is currently set. (1024 for system ulimit, and 8192 for Docker). These limits are rather weak, as for a normal ElasticSearch deployment the pods won't even start if nofiles < 65536.

Which service(s) is this request for?
EKS

Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?
In this case, I am trying to release elasticsearch. In the old KOPS cluster, the default limits were set much higher than in this case.

Are you currently working around this issue?
To work around the issue, I have to add userdata which manually changes ulimits in /etc/security/limits.conf for system-wide limits, and update /etc/docker/daemon.json to set limits for Docker.

[kamirendawkins]

I think this will be solved by awslabs/amazon-eks-ami#206

[rtkgjacobs]

Some deployments will also modify this per container / pod. This helm chart uses an init container to do this w/ Elasticsearch on K8s. I assume this approach below steps around the above issue until its corrected in future AMI's by default?

https://github.com/helm/charts/blob/master/stable/elasticsearch/templates/data-statefulset.yaml#L77

EDIT - wait sorry, I mixed up the setting meaning, ignore this comment above.

[trimbleAdam]

Closing as dupe of awslabs/amazon-eks-ami#206