#kubernetes (2023-06)
Archive: https://archive.sweetops.com/kubernetes/
2023-06-11
When using hostNetwork is there a recommended port range to use? I was thinking of using ports 49152 to 65535 (ChatGPT recommended this).
2023-06-12
So, ideas… our clusters like my clusters have dependencies, things that it requires to run.. Here is a short list:
VPC CNI CoreDNS EBS/EFS CSI ingress nginx AWSLoadBalancerController ClusterAutoScaler Deprecated APIversions Kube Proxy
All of these ‘things/dependencies’ require various versions of k8s, but that I mean
Kube-Proxy v1 needs k8s v1.5 (randomly made up example)
is anyone aware of an API/endpoint/something that maintains a matrix of all of this that can be queried?
thats what we are trying to determine.
First things you may need to have more insights into what the applications need from k8s, so you can create a newest version or -1 version of k8s testing environment to play with
CoreDNS and Kube Proxy can be replaced in the stack, Nginx ingress can be replaced by AWSLoadBalancerController
etc.
Deprecated APIversions is related to the applications
Forgot to mention this the other day but Pluto and a tool called Ent might be handy for your use case.
https://github.com/doitintl/kube-no-trouble
https://github.com/FairwindsOps/pluto
In theory you should be able to run kube no trouble and see if you are using any deprecated APIs
I use to Kubent and Pluto before my EKS upgrades but similar to @Jurgen I’m looking for a tool that can take inventory of which version of a particular app running e.g. ingress nginx running version 1.2.1 and therefore incompatible with 1.24. I’ve started scripting this out but it seems messy and inconsistent
2023-06-13
2023-06-14
2023-06-21
Hi @everyone, let’s assume I have an app that doesn’t have any health endpoint or port, now how can I decide whether the app is ready or not…can health probes use pod logs to determine the status?
yeah, so you write a simple Flask service only for health check endpoint and in the service it can check logs
then in k8s yaml, refer to the endpoint for health checks
Use an exec type - any command can be used as a probe.
There’s an example in the docs here: https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-a-liveness-command
But then I have to read the log file and the size of the log file keeps on increasing which may give timeout or fail in future.
I’m not sure how else you were expecting to use a log file…
You could have your app touch a file every x seconds and the command could check the last modified time. If it’s too old that means your app has failed, so you exit with a non-zero exit status. This would cause the probe to fail.
2023-06-22
Anyone used argoCD and also the Pulumi operator? Would love to read a little on your experience. ArgoCD seems more common but would love to know if anyone is benefiting from Pulumi in k8s in a way that argoCD doesn’t solve
2023-06-26
Posted some info on why we’ve decided to skip on Crossplane to the blog today: https://masterpoint.io/updates/passing-on-crossplane/. Would love to hear folks thoughts!
Great post, thanks for the details!
Thanx for sharing!
2023-06-29
https://medium.com/dev-genius/talking-about-the-sidecar-design-pattern-in-k8s-part-1-ab3aefca3a88 https://medium.com/@xpf6677/talking-about-the-sidecar-design-pattern-in-k8s-part-2-53dfd98bd679 Hi forks 
! I wrote a blog about Kubernetes sidecar design patterns and and would love to hear your feedback!