#aws - July 2024

Terrascan isn't properly identifying any of the cloudposse modules for compliance. Is there a scanner that works with cloudposse modules?

Hi All,
Seeking suggestions for a AWS POC,

Setting up a small AWS POC, planning to setup 1uat machine 1 prod machine and 1 Jenkins machine to build and deploy to both uat and prod.

To ensure the security, planning to go with aws organisation and keep 3accounts to keep all 3 servers. Is it good approach or any other approach to set it up? interms of security and cost effective.

Thanks in advance. 🙂

Hi Everyone,
need help in python runtime upgrade in aws lambda,
I have deployed datadog as aws lambda application with python runtime as 3.7 a while ago.
Have a lot of env vars in it. How do we upgrade the application with python 3.11 runtime.
thanks in advance.

I did try by just manually upgrading the lambda function runtime to python3.11 but it breaks

I have a ALB in a source acct routing to a NLB in a target account at the moment ...we have a use case to have Private Link setup from another source account ...can the endpont link be setup with the same NLB in target account by creating endpoint service.. i want both routes to work

Hey all, curious what you all think.

Jupyterhub Notebooks on EKS has a worst case scenario cold start where a Data Scientist needs to wait for a Node to spin up and for the large Docker Image to pull.

The thinking is that we can largely eliminate (or at least reduce) the Docker Image pull time by creating AMIs with the Docker Image on them (with Image Builder pulled as ec2-user). Jupyterhub would then launch workloads (notebook servers) onto these AMIs as Nodes managed by Karpenter with Taints/Tolerations and Node Affinity.

However, it seems like ec2-user and the kubelet (or containerd?) have different docker storage (there's only one EBS volume attached). This is causing EKS to pull images that should already be available to it because the image was previously pulled by ec2-user.

Running a docker images command run on the node (via SSH as ec2-user) shows a couple our latest tag which was pulled while building the AMI. Launching a Notebook with a specific tag "foo" caused a docker pull to occur. When it was finished, running docker images via SSH again did not show foo in the output.

Conversely, pulling a different tag bar as ec2-user and then launching a Notebook Server with bar caused EKS to pull the Image again.

Any ideas?

Hey guys, I see that CloudPosse prefers using Ecspresso as the ECS cli tooling. I'm curious to hear why that's the case, and what do you look for? - and if there's benefits y'all see in using this versues other tools like ecs-deploy , or even plain AWS commands as a script as as this https://github.com/silinternational/ecs-deploy/tree/develop? From what I'm seeing, ecspresso definitely has better task definition control.

any body here have an idea on how to get the live data of dynamodb?
planning to stream the data from dynamodb to bigquery or dynamodb to s3 to bigquery.

any insights or idea without using a third party tool. thank you so much.

Hi hi, anyone here uses prometheus thanos grafana stack? I have 4 AWS envs (different accounts) and want to set up prometheus in all these envs but only one thanos and grafana. Trying to see what the industry standard is for connecting it all together. Vpc peering or transit gateways seem insecure.