Kubernetes

🌐 Network troubleshooting
🌐 Network troubleshooting
Troubleshoot DNS vi dns.yml 1apiVersion: v1 2kind: Pod 3metadata: 4 name: dnsutils 5 namespace: default 6spec: 7 containers: 8 - name: dnsutils 9 image: registry.k8s.io/e2e-test-images/jessie-dnsutils:1.3 10 command: 11 - sleep 12 - "infinity" 13 imagePullPolicy: IfNotPresent 14 restartPolicy: Always deploy dnsutils 1k apply -f dns.yml 2pod/dnsutils created 3 4kubectl get pods dnsutils 5NAME READY STATUS RESTARTS AGE 6dnsutils 1/1 Running 0 36s Troubleshoot with dnsutils 1kubectl exec -i -t dnsutils -- nslookup kubernetes.default 2;; connection timed out; no servers could be reached 3command terminated with exit code 1 4 5kubectl exec -ti dnsutils -- cat /etc/resolv.conf 6search default.svc.cluster.local svc.cluster.local cluster.local example.local 7nameserver 10.43.0.10 8options ndots:5 9 10kubectl get endpoints kube-dns --namespace=kube-system 11NAME ENDPOINTS AGE 12kube-dns 10.42.0.6:53,10.42.0.6:53,10.42.0.6:9153 5d1h 13 14kubectl get svc kube-dns --namespace=kube-system 15NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE 16kube-dns ClusterIP 10.43.0.10 <none> 53/UDP,53/TCP,9153/TCP 5d1h CURL 1cat << EOF > curl.yml 2apiVersion: v1 3kind: Pod 4metadata: 5 name: curl 6 namespace: default 7spec: 8 containers: 9 - name: curl 10 image: curlimages/curl 11 command: 12 - sleep 13 - "infinity" 14 imagePullPolicy: IfNotPresent 15 restartPolicy: Always 16EOF 17 18k apply -f curl.yml 19 20#Test du DNS 21kubectl exec -i -t curl -- curl -v telnet://10.43.0.10:53 22kubectl exec -i -t curl -- curl -v telnet://kube-dns.kube-system.svc.cluster.local:53 23kubectl exec -i -t curl -- nslookup kube-dns.kube-system.svc.cluster.local 24 25curl -k -I --resolve subdomain.domain.com:52.165.230.62 https:/subdomain.domain.com/
🎡 Helm
🎡 Helm
Administration See what is currently installed 1helm list -A 2NAME NAMESPACE REVISION UPDATED STATUS CHART APP VERSION 3nesux3 default 1 2022-08-12 20:01:16.0982324 +0200 CEST deployed nexus3-1.0.6 3.37.3 Install/Uninstall 1helm status nesux3 2helm uninstall nesux3 3helm install nexus3 <chart> # chart URL or path 4helm history nexus3 5 6# work even if already installed 7helm upgrade --install ingress-nginx ${DIR}/helm/ingress-nginx \ 8 --namespace=ingress-nginx \ 9 --create-namespace \ 10 -f ${DIR}/helm/ingress-values.yml 11 12#Make helm unsee an apps (it does not delete the apps) 13kubectl delete secret -l owner=helm,name=argo-cd Handle Helm Repo and Charts 1#Handle repo 2helm repo list 3helm repo add gitlab https://charts.gitlab.io/ 4helm repo update 5 6#Pretty usefull to configure 7helm show values elastic/eck-operator 8helm show values grafana/grafana --version 8.5.1 9 10#See different version available 11helm search repo hashicorp/vault 12helm search repo hashicorp/vault -l 13 14# download a chart 15helm fetch ingress/ingress-nginx --untar
🎲 Kubectl
🎲 Kubectl
Connection to k8s cluster Kubeconfig Define KUBECONFIG in your profile 1# Default one 2KUBECONFIG=~/.kube/config 3 4# Several context - to keep splited 5KUBECONFIG=~/.kube/k3sup-lab:~/.kube/k3s-dev 6 7# Or can be specified in command 8kubectl get pods --kubeconfig=admin-kube-config View and Set 1kubectl config view 2kubectl config current-context 3 4kubectl config set-context \ 5dev-context \ 6--namespace=dev-namespace \ 7--cluster=docker-desktop \ 8--user=dev-user 9 10kubectl config use-context lab Switch context 1#set Namespace 2kubectl config set-context --current --namespace=nexus3 3kubectl config get-contexts Kubecm The problem with the kubeconfig is that it get nexted in one kubeconfig and difficult to manage on long term. The best way to install it, is with Arkade arkade get kubecm - see arkade.
🐄 RKE2
🐄 RKE2
General Checks Nice gist to troubleshoot etcd link 1journalctl -u rke2-server.service -f 2 3tail -f /var/lib/rancher/rke2/agent/containerd/containerd.log 4 5tail -f /var/lib/rancher/rke2/agent/logs/kubelet.log 6 7# crictl 8export CRI_CONFIG_FILE=/var/lib/rancher/rke2/agent/etc/crictl.yaml 9/var/lib/rancher/rke2/bin/crictl ps 10 11/var/lib/rancher/rke2/bin/crictl --config /var/lib/rancher/rke2/agent/etc/crictl.yaml ps 12 13/var/lib/rancher/rke2/bin/crictl --runtime-endpoint unix:///run/k3s/containerd/containerd.sock ps -a 14 15/var/lib/rancher/rke2/bin/ctr --address /run/k3s/containerd/containerd.sock --namespace k8s.io container ls 16 17# Kubectl 18export KUBECONFIG=/etc/rancher/rke2/rke2.yaml 19export PATH=$PATH:/usr/local/bin/:/var/lib/rancher/rke2/bin/ 20kubectl get addon -A Check etcd endpoint status 1export CRI_CONFIG_FILE=/var/lib/rancher/rke2/agent/etc/crictl.yaml 2etcdcontainer=$(/var/lib/rancher/rke2/bin/crictl ps --label io.kubernetes.container.name=etcd --quiet) 3/var/lib/rancher/rke2/bin/crictl exec $etcdcontainer sh -c "ETCDCTL_ENDPOINTS='https://127.0.0.1:2379' ETCDCTL_CACERT='/var/lib/rancher/rke2/server/tls/etcd/server-ca.crt' ETCDCTL_CERT='/var/lib/rancher/rke2/server/tls/etcd/server-client.crt' ETCDCTL_KEY='/var/lib/rancher/rke2/server/tls/etcd/server-client.key' ETCDCTL_API=3 etcdctl endpoint status --cluster --write-out=table" Check etcd health status 1export CRI_CONFIG_FILE=/var/lib/rancher/rke2/agent/etc/crictl.yaml 2etcdcontainer=$(/var/lib/rancher/rke2/bin/crictl ps --label io.kubernetes.container.name=etcd --quiet) 3/var/lib/rancher/rke2/bin/crictl exec $etcdcontainer sh -c "ETCDCTL_ENDPOINTS='https://127.0.0.1:2379' ETCDCTL_CACERT='/var/lib/rancher/rke2/server/tls/etcd/server-ca.crt' ETCDCTL_CERT='/var/lib/rancher/rke2/server/tls/etcd/server-client.crt' ETCDCTL_KEY='/var/lib/rancher/rke2/server/tls/etcd/server-client.key' ETCDCTL_API=3 etcdctl endpoint health --cluster --write-out=table"
🐎 K3D
🐎 K3D
K3D equal k3s in a container. a tools to create single- and multi-node k3s clusters. Our favorite use case, is with podman and rootless. So there is some customization upstream to do. One downside I’ve found with k3d is that the Kubernetes version it uses is behind the current k3s release. Note for ARM PC: 1sudo apt install qemu-user-static 2podman run --rm --privileged multiarch/qemu-user-static --reset -p yes Install 1# Manual way 2curl -s https://raw.githubusercontent.com/k3d-io/k3d/main/install.sh | bash 3 4# or with arkade: 5arkade get k3d 6 7# Auto-completion 8k3d completion zsh > "$ZSH/completions/_k3d" Tweaks for podman and rootless The issue: 1k3d cluster create test 2 3ERRO[0000] Failed to get nodes for cluster 'test': docker failed to get containers with labels 'map[k3d.cluster:test]': failed to list containers: permission denied while trying to connect to the Docker daemon socket at unix:///var/run/docker.sock: Get "http://%2Fvar%2Frun%2Fdocker.sock/v1.46/containers/json?all=1&filters=%7B%22label%22%3A%7B%22app%3Dk3d%22%3Atrue%2C%22k3d.cluster%3Dtest%22%3Atrue%7D%7D": dial unix /var/run/docker.sock: connect: permission denied The solution: 1# TODO 2loginctl enable-linger $(whoami) 3 4# Either reload terminal or do below: 5export XDG_RUNTIME_DIR=/tmp/run-$(id -u) 6mkdir -p $XDG_RUNTIME_DIR 7chmod 700 $XDG_RUNTIME_DIR 8 9sudo mkdir -p /etc/containers/containers.conf.d 10sudo sh -c "echo 'service_timeout=0' > /etc/containers/containers.conf.d/timeout.conf" 11 12sudo ln -s /run/podman/podman.sock /var/run/docker.sock 13 14XDG_RUNTIME_DIR=${XDG_RUNTIME_DIR:-/run/user/$(id -u)} 15export DOCKER_HOST=unix://$XDG_RUNTIME_DIR/podman/podman.sock 16export DOCKER_SOCK=$XDG_RUNTIME_DIR/podman/podman.sock 17 18systemctl --user enable --now podman.socket If /sys/fs/cgroup/cgroup.controllers is present on your system, you are using v2, otherwise you are using v1.
🐙 ArgoCD
🐙 ArgoCD
What is ArgoCD ArgoCD is a declarative, GitOps continuous delivery tool for Kubernetes. Your Git repository is the single source of truth; ArgoCD watches it and continuously syncs the cluster to match the desired state. 1Git repo ──► ArgoCD ──► Kubernetes cluster Key concepts Application — a group of Kubernetes resources described in Git. Project — groups applications and scopes their permissions. Source — the repo to render from (git, helm, kustomize, …). Sync — reconcile the live cluster with the desired state. Install 1kubectl create namespace argocd 2kubectl apply -n argocd \ 3 -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml 4 5# expose the UI 6kubectl port-forward svc/argocd-server -n argocd 8080:443 CLI 1argocd login localhost:8080 2 3argocd app create myapp \ 4 --repo https://github.com/org/repo.git \ 5 --path deploy \ 6 --dest-server https://kubernetes.default.svc \ 7 --dest-namespace default 8 9argocd app list 10argocd app sync myapp 11argocd app get myapp Ops notes With syncPolicy.automated set, ArgoCD re-applies any drift on its own. Don’t commit plaintext secrets to Git — pair ArgoCD with an external secret manager (Vault, sealed-secrets, SOPS/ksops).
🐠 OKD & OpenShift
🐠 OKD & OpenShift
OKD vs OpenShift They are essentially the same Kubernetes distribution. OKD — short for “The Community Distribution of Kubernetes that powers Red Hat OpenShift” — is the free, upstream/community edition: the same codebase, community support, and images pulled from quay.io. OpenShift is Red Hat’s enterprise product: OKD plus commercial support, certifications, a longer support lifecycle, Red Hat registries (registry.redhat.io) and OperatorHub access. Day-to-day they are nearly interchangeable — both use the same oc, openshift-install and oc-mirror tooling, and the same install-config.yaml layout. Unless stated otherwise, the notes below apply to both.
🐣 Bash Functions for k8s
🐣 Bash Functions for k8s
A list of nice findings for Kubernetes List all images in Helm chart 1images=$(helm template $helm |yq -N '..|.image? | select(. == "*" and . != null)'|sort|uniq|grep ":"|egrep -v '*:[[:blank:]]' || echo "") upload images listed in an Helm chart 1load_helm_images(){ 2 # look in helm charts 3 for helm in $(ls ../../roles/*/files/helm/*.tgz); do 4 printf "\e[1;34m[INFO]\e[m Look for images in ${helm}...\n" 5 6 images=$(helm template $helm |yq -N '..|.image? | select(. == "*" and . != null)'|sort|uniq|grep ":"|egrep -v '*:[[:blank:]]' || echo "") 7 8 dir=$( dirname $helm | xargs dirname ) 9 10 echo "####" 11 12 if [ "$images" != "" ]; then 13 printf "\e[1;34m[INFO]\e[m Images found in the helm charts: ${images}\n" 14 printf "\e[1;34m[INFO]\e[m Create ${dir}/images images...\n" 15 16 mkdir -p ${dir}/images 17 18 while i= read -r image_name; do 19 archive_name=$(basename -a $(awk -F : '{print $1}'<<<${image_name})); 20 printf "\e[1;34m[INFO]\e[m Pull images...\n" 21 podman pull ${image_name}; 22 printf "\e[1;34m[INFO]\e[m Push ${image_name} in ${dir}/images/${archive_name}\n" 23 podman save ${image_name} --format oci-archive -o ${dir}/images/${archive_name}; 24 done <<< ${images} 25 else 26 printf "\e[1;34m[INFO]\e[m No Images found in the helm charts: $helm\n" 27 fi 28 done 29} Check components version 1function checkComponentsInstall() { 2 componentsArray=("kubectl" "helm") 3 for i in "${componentsArray[@]}"; do 4 command -v "${i}" >/dev/null 2>&1 || 5 { echo "[ERROR] ${i} is required, but it's not installed. Aborting." >&2; exit 1; } 6 done 7} Version comparator 1function checkK8sVersion() { 2 currentK8sVersion=$(kubectl version --short | grep "Server Version" | awk '{gsub(/v/,$5)}1 {print $3}') 3 testVersionComparator 1.20 "$currentK8sVersion" '<' 4 if [[ $k8sVersion == "ok" ]]; then 5 echo "current kubernetes version is ok" 6 else 7 minikube start --kubernetes-version=v1.22.4; 8 fi 9} 10 11 12# the comparator based on https://stackoverflow.com/a/4025065 13versionComparator () { 14 if [[ $1 == $2 ]] 15 then 16 return 0 17 fi 18 local IFS=. 19 local i ver1=($1) ver2=($2) 20 # fill empty fields in ver1 with zeros 21 for ((i=${#ver1[@]}; i<${#ver2[@]}; i++)) 22 do 23 ver1[i]=0 24 done 25 for ((i=0; i<${#ver1[@]}; i++)) 26 do 27 if [[ -z ${ver2[i]} ]] 28 then 29 # fill empty fields in ver2 with zeros 30 ver2[i]=0 31 fi 32 if ((10#${ver1[i]} > 10#${ver2[i]})) 33 then 34 return 1 35 fi 36 if ((10#${ver1[i]} < 10#${ver2[i]})) 37 then 38 return 2 39 fi 40 done 41 return 0 42} 43 44testVersionComparator () { 45 versionComparator $1 $2 46 case $? in 47 0) op='=';; 48 1) op='>';; 49 2) op='<';; 50 esac 51 if [[ $op != "$3" ]] 52 then 53 echo "Kubernetes test fail: Expected '$3', Actual '$op', Arg1 '$1', Arg2 '$2'" 54 k8sVersion="not ok" 55 else 56 echo "Kubernetes test pass: '$1 $op $2'" 57 k8sVersion="ok" 58 fi 59}
📜 CertManager
📜 CertManager
What is Cert-Manager cert-manager automates the management of X.509 certificates inside Kubernetes via CRDs — it requests, issues, renews and rotates certificates automatically. Key concepts Issuer — a namespaced certificate signer. ClusterIssuer — a cluster-wide signer. Certificate — asks cert-manager to obtain a cert for a name. Supported backends: ACME (Let’s Encrypt), self-signed, CA, Vault, Venafi, … Install (helm) 1helm repo add jetstack https://charts.jetstack.io 2helm repo update 3helm upgrade --install cert-manager jetstack/cert-manager \ 4 --namespace cert-manager --create-namespace \ 5 --set installCRDs=true Example: self-signed issuer 1apiVersion: cert-manager.io/v1 2kind: ClusterIssuer 3metadata: 4 name: selfsigned 5spec: 6 selfSigned: {} 1apiVersion: cert-manager.io/v1 2kind: Certificate 3metadata: 4 name: example-tls 5spec: 6 secretName: example-tls 7 dnsNames: 8 - example.com 9 issuerRef: 10 name: selfsigned 11 kind: ClusterIssuer Useful commands 1kubectl get certificate -A 2kubectl get certificaterequest -A 3kubectl describe certificate <name> -n <ns> 4 5# manual renewal (normally automatic) 6cmctl renew <name> -n <ns> Ops notes Cert-Manager stores the TLS key/cert in a <name> secret, ready for Ingress. kubectl get challenges is the first place to look for ACME/Let’s Encrypt failures.
🔒 Vault on k8s
🔒 Vault on k8s
Some time ago, I made a small shell script to handle Vault on a cluster kubernetes. For documentation purpose. Install Vault with helm 1#!/bin/bash 2 3## Variables 4DIRNAME=$(dirname $0) 5DEFAULT_VALUE="vault/values-override.yaml" 6NewAdminPasswd="PASSWORD" 7PRIVATE_REGISTRY_USER="registry-admin" 8PRIVATE_REGISTRY_PASSWORD="PASSWORD" 9PRIVATE_REGISTRY_ADDRESS="registry.example.com" 10DOMAIN="example.com" 11INGRESS="vault.${DOMAIN}" 12 13if [ -z ${CM_NS+x} ];then 14 CM_NS='your-namespace' 15fi 16 17if [ -z ${1+x} ]; then 18 VALUES_FILE="${DIRNAME}/${DEFAULT_VALUE}" 19 echo -e "\n[INFO] Using default values file '${DEFAULT_VALUE}'" 20else 21 if [ -f $1 ]; then 22 echo -e "\n[INFO] Using values file $1" 23 VALUES_FILE=$1 24 else 25 echo -e "\n[ERROR] No file exist $1" 26 exit 1 27 fi 28fi 29 30## Functions 31function checkComponentsInstall() { 32 componentsArray=("kubectl" "helm") 33 for i in "${componentsArray[@]}"; do 34 command -v "${i}" >/dev/null 2>&1 || 35 { echo "${i} is required, but it's not installed. Aborting." >&2; exit 1; } 36 done 37} 38 39function createSecret() { 40kubectl get secret -n ${CM_NS} registry-pull-secret --no-headers 2> /dev/null \ 41|| \ 42kubectl create secret docker-registry -n ${CM_NS} registry-pull-secret \ 43 --docker-server=${PRIVATE_REGISTRY_ADDRESS} \ 44 --docker-username=${PRIVATE_REGISTRY_USER} \ 45 --docker-password=${PRIVATE_REGISTRY_ADDRESS} 46} 47 48function installWithHelm() { 49helm dep update ${DIRNAME}/helm 50 51helm upgrade --install vault ${DIRNAME}/helm \ 52--namespace=${CM_NS} --create-namespace \ 53--set global.imagePullSecrets.[0]=registry-pull-secret \ 54--set global.image.repository=${PRIVATE_REGISTRY_ADDRESS}/hashicorp/vault-k8s \ 55--set global.agentImage.repository=${PRIVATE_REGISTRY_ADDRESS}/hashicorp/vault \ 56--set ingress.hosts.[0]=${INGRESS} \ 57--set ingress.enabled=true \ 58--set global.leaderElection.namespace=${CM_NS} 59 60echo -e "\n[INFO] sleep 30s" && sleep 30 61} 62 63checkComponentsInstall 64createSecret 65installWithHelm Init Vault on kubernetes Allow local kubernetes to create and reach secret on the Vault
🔱 K3S
🔱 K3S
Specific to RHEL 1# Create a trust zone for the two interconnect 2sudo firewall-cmd --permanent --zone=trusted --add-source=10.42.0.0/16 #pods 3sudo firewall-cmd --permanent --zone=trusted --add-source=10.43.0.0/16 #services 4sudo firewall-cmd --reload 5sudo firewall-cmd --list-all-zones 6 7# on Master 8sudo rm -f /var/lib/cni/networks/cbr0/lock 9sudo /usr/local/bin/k3s-killall.sh 10sudo systemctl restart k3s 11sudo systemctl status k3s 12 13# on Worker 14sudo rm -f /var/lib/cni/networks/cbr0/lock 15sudo /usr/local/bin/k3s-killall.sh 16sudo systemctl restart k3s-agent 17sudo systemctl status k3s-agent Check Certificates 1# Get CA from K3s master 2openssl s_client -connect localhost:6443 -showcerts < /dev/null 2>&1 | openssl x509 -noout -enddate 3openssl s_client -showcerts -connect 193.168.51.103:6443 < /dev/null 2>/dev/null|openssl x509 -outform PEM 4openssl s_client -showcerts -connect 193.168.51.103:6443 < /dev/null 2>/dev/null|openssl x509 -outform PEM | base64 | tr -d '\n' 5 6# Check end date: 7for i in `ls /var/lib/rancher/k3s/server/tls/*.crt`; do echo $i; openssl x509 -enddate -noout -in $i; done 8 9# More efficient: 10cd /var/lib/rancher/k3s/server/tls/ 11for crt in *.crt; do printf '%s: %s\n' "$(date --date="$(openssl x509 -enddate -noout -in "$crt"|cut -d= -f 2)" --iso-8601)" "$crt"; done | sort 12 13# Check CA issuer 14for i in $(find . -maxdepth 1 -type f -name "*.crt"); do openssl x509 -in ${i} -noout -issuer; done Rancher 1# Rancher local install - for example on WSL 2sudo podman run --privileged -d --restart=unless-stopped -p 80:80 -p 443:443 rancher/rancher 3sudo podman ps 4sudo podman logs 74533d50d991 2>&1 | grep "Bootstrap Password:"
🚀 Operator SDK
🚀 Operator SDK
What is an Operator A controller is a loop that watches Kubernetes objects and reconciles their desired state (the spec) with the actual state (the cluster). An Operator is a controller with operational knowledge baked in: it knows how to install, configure, scale, back up and upgrade an application — the jobs a human admin used to do by hand. 1desired state (CR) ──► reconcile loop ──► actual state (cluster) Classic examples: etcd-operator, prometheus-operator.