amazon-eks-ami VS conduit

Careful assuming system reserved will be present. Last I checked, AWS EKS does not have system reserved resources for the kubelet by default and as a result, pods can starve those for resources (e.g., https://github.com/awslabs/amazon-eks-ami/issues/79). This is of course more important for memory, but could impact CPU as well.

For example, this is available for AL2: https://github.com/awslabs/amazon-eks-ami

I found this document that lists the pod limits per node size. I suspect you will want to consider larger worker nodes or you will very quickly be unable to schedule additional workloads.

source

Issue 1 is a known issue due to memory reservation being to low, see e.g. https://github.com/awslabs/amazon-eks-ami/issues/1145

No, the new node is added if there are not enough resiurces to start a new pod. So if you have many pods with small resource usage you can hit the pod per node limit, on eks you have a max number of pods depending on the instance type - https://github.com/awslabs/amazon-eks-ami/blob/master/files/eni-max-pods.txt You can incerase that limit : https://docs.aws.amazon.com/eks/latest/userguide/cni-increase-ip-addresses.html

# of pods are essentially capped by the worker node choice.

below excerpt from: https://github.com/awslabs/amazon-eks-ami/blob/master/files/...

  # Mapping is calculated from AWS EC2 API using the following formula:

See also: EKS nodes lose readiness when containers exhaust memory

So it manifests itself in this way: your pod is scheduled but remains pending forever. You check the logs and you see that it's complaining that the an IP address. Ultimately, if you check here, you see the maximum number of pods that can be scheduled on any underlying ec2 instance, even if you have remaining IPs in your subnet. I found this to be one of the most poorly understood phenomena in EKS. Even those who claimed to "crack" it and wrote fancy blog posts about it fundamentally got it wrong. AFAIK this document reflects the official AWS guide on how to mitigate this.

Leverage a service mesh like Istio or Linkerd to manage communication between microservices within the Kubernetes cluster. These service meshes can be configured to intercept JMX traffic and enforce access control policies. Benefits:

Looks like CNCF waved them through Graduation anyway, let's look at policies from July 28, 2021 when they were deemed "Graduated"

All maintainers of the LinkerD project had @boyant.io email addresses. [0] They do list 4 other members of a "Steering Committee", but LinkerD's GOVERNANCE.md gives all of the power to maintainers: [1]

> Ideally, all project decisions are resolved by maintainer consensus. If this is not possible, maintainers may call a vote. The voting process is a simple majority in which each maintainer receives one vote.

And CNCF Graduation policy says a project must "Have committers from at least two organizations" [2]. So it appears that the CNCF accepted the "Steering Committee" as an acceptable 2nd committer, even though the Governance policy still gave the maintainers all of the power.

I would like to know if the Steering Committee voted to remove stable releases from an un-biased position acting in the best interest of the project, or if they were simply ignored or not even advised on the decision.

I'm all for Boyant doing what they need to do to make money and survive as a Company. But at that point my opinion is that they should withdraw the project from the CNCF and stop pretending like the foundation has any influence on the project's governance.

[0] https://github.com/linkerd/linkerd2/blob/489ca1e3189b6a5289d...

From here, we can explore other developments and tutorials on Kubernetes, such as o11y or observability (PLG, ELK, ELF, TICK, Jaeger, Pyroscope), service mesh (Linkerd, Istio, NSM, Consul Connect, Cillium), and progressive delivery (ArgoCD, FluxCD, Spinnaker).

https://linkerd.io/ is a much lighter-weight alternative but you do still get some of the fancy things like mtls without needing any manual configuration. Install it, label your namespaces, and let it do it's thing!

Would it be possible to create a custom extension with the code that authorize traffic based on my custom access token?

Open source API Gateway (Apache APISIX and Traefik), Service Mesh (Istio and Linkerd) solutions are capable of doing traffic splitting and implementing functionalities like Canary Release and Blue-Green deployment. With canary testing, you can make a critical examination of a new release of an API by selecting only a small portion of your user base. We will cover the canary release next section.

I have experimented with other service meshes and I was able to get up to speed quickly: Linkerd = 1 day, Istio = 3 days, NGINX Service Mesh = 5 days, but Consul Connect service mesh took at least 11 days to get off the ground. This is by far the most complex solution available.

https://github.com/linkerd/linkerd2 (random example)

Out of the number of service mesh solutions that exist, the most popular open source ones are: Linkerd, Istio, and Consul. Here at Koyeb, we are using Kuma.