cilium-kind-lab

This repo creates two local kind clusters and uses Cilium as their CNI.

I'm trying to use Cilium as much as possible to implement Service Mesh, Cluster Mesh, Gateway API, monitoring and observability

For now, these cluster hold no particular application.

The environment

The following tools are required in order to run the init script:

• docker (will NOT be installed by script)
• helm (will NOT be installed by script)
• kind (will NOT be installed by script)
• kubectl (will NOT be installed by script)
• cilium (will be installed by script)
• hubble (will be installed by script)

Please note that kind cluster may fail to start because of low resources limits: check https://kind.sigs.k8s.io/docs/user/known-issues/#pod-errors-due-to-too-many-open-files.

My current inotify limits:

sysctl -a
...
fs.inotify.max_queued_events = 16384
fs.inotify.max_user_instances = 1280
fs.inotify.max_user_watches = 655360
...

Following my working environment where I developed this repo:

$ lsb_release -a
No LSB modules are available.
Distributor ID:	Linuxmint
Description:	Linux Mint 21
Release:	21
Codename:	vanessa

$ uname -r
6.8.0-40-generic

$ docker version -f json | jq '.Client.Version'
"27.4.1"

$ helm version --short
v3.12.0+gc9f554d

$ kind version
kind v0.26.0 go1.23.4 linux/amd64

$ kubectl version --client -o json | jq '.clientVersion.gitVersion'
"v1.29.0"

$ cilium version
cilium-cli: v0.16.22 compiled with go1.23.4 on linux/amd64
cilium image (default): v1.16.4
cilium image (stable): v1.16.5
cilium image (running): 1.16.5

$ hubble version
hubble v1.16.5@HEAD-6dbbd44 compiled with go1.23.4 on linux/amd64

The clusters

Each cluster is made of 1 control-plane node and 2 worker nodes.

In order to create the clusters, just run ./setup.sh.

$ ./setup.sh

TODO

• replace kube-proxy with Cilium
• create more than one kind cluster and make them communicate via Cilium Cluster Mesh
• install k8s dashboard
• install Hubble
• enable Cilium Service Mesh
• expose Services by using Cilium Gateway API