Kubeletter

Monitor k8s cluster by executing kubectl commands

How it works

1. execute commands like kubectl top nodes
2. parse the printed results and process into data
3. compare the data with past data and make report out of it
4. deliver the report via Slack and email

Prerequisites

configure kubectl to connect to your k8s cluster

this is already ready in k8s containers in most cases

How to run

Use your local kubectl to develop

proxy

$ kubectl proxy --address="0.0.0.0" --accept-hosts '.*' --reject-methods=NONE

Example ENVs

if you use docker for mac

KUBERNETES_SERVICE_PORT=8001
KUBERNETES_SERVICE_HOST=docker.for.mac.localhost
INDIVISUAL_COUNT_LIMIT=3 # default is 5 and -1 means no limit

Running as a k8s container

Simply pass just RUNNING_INSIDE_K8S=true to run kubectl properly inside a k8s cluster

No need to provide kube related env inside k8s cluster to execute kubectl

Troubleshooting

It seems like it's not working inside k8s cluster

you can debug this by entering the pod and run necessaray commands like below

• kubectl get pod
• kubectl get node
• kubectl get hpa
• kubectl top node
• kubectl top pod
• kubectl cluster-info

if any of commands above doesn't work, that's probably because your service account tied to your deployment has not enough permissions; in that case you might want to look at https://kubernetes.io/docs/reference/access-authn-authz/rbac/