Kubernetes Scripts

A collection of scripts for various tasks in Kubernetes.

Usage

Each script has a usage function. See usage with

$ <script> --help

Scripts

findEmptyNamespaces.sh: Loop over all namespaces in a cluster and find empty ones.
getPodsTopCSV.sh: Get a pod's cpu and memory usage (optionally per container) written as CSV formatted file.
getResourcesCSV.sh: Get all pods resources requests and limits per container in a CSV format with values normalized. CSV format is very automation friendly and is great for pasting in an Excel or Google sheet for further processing.
getRestartingPods.sh: Get all pods (all or single namespace) that have restarts detected in one or more containers. Formatted in CSV.
podReady: Simple script to check if pod is really ready. Check status is 'Running' and that all containers are ready. Returns 0 if ready. Returns 1 if not ready.

YAML

podWithHostFS.yaml: A pod with host root file system mounted into it.
WARNING: There is danger of corrupting your Kubernetes host. Use with extra care!

One liners

Kubectl

See all custer nodes load (top)

kubectl top nodes

Get cluster events

# All cluster
kubectl get events

# Specific namespace events
kubectl get events --namespace=kube-system

Get all cluster nodes IPs and names

# Single call to K8s API
kubectl get nodes -o json | grep -A 12 addresses

# A loop for more flexibility
for n in $(kubectl get nodes -o name); do \
  echo -e "\nNode ${n}"; \
  kubectl get ${n} -o json | grep -A 8 addresses; \
done

See all cluster nodes CPU and Memory requests and limits

kubectl describe nodes | grep -A 3 "Name:\|Resource .*Requests .*Limits" | grep -v "Roles:"

Get all labels attached to all pods in a namespace

for a in $(kubectl get pods -n namespace1 -o name); do \
  echo -e "\nPod ${a}"; \
  kubectl -n namespace1 describe ${a} | awk '/Labels:/,/Annotations/' | sed '/Annotations/d'; \
done

Forward local port to a pod or service

# Forward localhost port 8080 to a specific pod exposing port 8080
kubectl port-forward -n namespace1 web 8080:8080

# Forward localhost port 8080 to a specific web service exposing port 80
kubectl port-forward -n namespace1 svc/web 8080:80

Port forwarding

A great tool for port forwarding all services in a namespace + adding aliases to /etc/hosts is kubefwd. Note that this requires root or sudo to allow temporary editing of /etc/host.

# Port forward all service in namespace1
kubefwd svc -n namespace1

Extract and decode a secret's value

# Get the value of the postgresql password
kubectl get secret -n namespace1 my-postgresql -o jsonpath="{.data.postgres-password}" | base64 --decode

Copy secret from `namespace1` to `namespace2`

kubectl get secret my-secret --namespace namespace1 -o yaml | sed "/namespace:/d" | kubectl apply --namespace=namespace2 -f -

Create an Ubuntu pod

A one liner to create an Ubuntu pod that will just wait forever.

# Create the pod
cat <<ZZZ | kubectl apply -f -
apiVersion: v1
kind: Pod
metadata:
  name: my-ubuntu-pod
spec:
  containers:
  - name: my-ubuntu-container
    image: ubuntu:20.04
    command:
    - 'bash'
    - '-c'
    - 'while true; do sleep 5; done'
ZZZ

# Shell into the pod
kubectl exec -it my-ubuntu-pod bash

# Delete the pods once done
kubectl delete pod my-ubuntu-pod

Start a shell in a temporary pod

Note - Pod will terminate once exited

# Ubuntu
kubectl run my-ubuntu --rm -i -t --restart=Never --image ubuntu -- bash

# CentOS
kubectl run my-centos --rm -i -t --restart=Never --image centos:8 -- bash

# Alpine
kubectl run my-alpine --rm -i -t --restart=Never --image alpine:3.10 -- sh

# Busybox
kubectl run my-busybox --rm -i -t --restart=Never --image busybox -- sh

Get formatted list of container images in pods

Useful for listing all running containers in your cluster

kubectl get pod -A -o=jsonpath='{range .items[*]}{.metadata.namespace},{.metadata.name},{.spec.containers[*].image}{"\n"}' | tr -s ' ' '\n'

Look into a few more examples of listing containers

Get list of pods sorted by restart count

Option 1 for all pods (Taken from kubectl cheatsheet)

kubectl get pods -A --sort-by='.status.containerStatuses[0].restartCount'

Option 2 with a filter and a CSV friendly output

kubectl get pods -A | grep my-app | awk '{print $5 ", " $1 ", " $6}'  | sort -n -r

Get current replica count on all HPAs (Horizontal Pod Autoscaler)

kubectl get hpa -A -o=custom-columns=NAME:.metadata.name,REPLICAS:.status.currentReplicas | sort -k2 -n -r

List non-running pods

kubectl get pods -A --field-selector=status.phase!=Running | grep -v Complete

Top Pods CPU or memory usage

# Top 20 pods by highest CPU usage
kubectl top pods -A | sort --reverse --key 3 --numeric | head -20

# Top 20 pods by highest memory usage
kubectl top pods -A | sort --reverse --key 4 --numeric | head -20

Helm

NOTE: It is recommended to move to Helm v3, which does not use tiller anymore.

Helm v2 on an RBAC enabled cluster

This will give tiller cluster-admin role

kubectl -n kube-system create sa tiller && \
    kubectl create clusterrolebinding tiller --clusterrole cluster-admin --serviceaccount=kube-system:tiller && \
    helm init --service-account tiller

Helm template

View the templates generated by helm install. Useful for seeing the actual templates generated by helm before deploying.
Can also be used for deploying the templates generated when cannot use Tiller

helm template <chart>

Debug helm install

Debug a helm install. Useful for seeing the actual values resolved by helm before deploying

helm install --debug --dry-run <chart>

Rolling restart

Roll a restart across all resources managed by a Deployment, DaemonSet or StatefulSet with zero downtime

# Deployment
kubectl -n <namespace> rollout restart deployment <deployment-name>

# DaemonSet
kubectl -n <namespace> rollout restart daemonset <daemonset-name>

# StatefulSet
kubectl -n <namespace> rollout restart statefulsets <statefulset-name>

Resources

Most of the code above is self experimenting and reading the docs. Some are copied and modified to my needs from other resources...

akria18/kubernetes-scripts