CKA-exam-5Feb-2023
Camille
  2023-02-04 11:18:40 2023-02-04 11:18      542 Words  3 Mins

View Resources

  • k api-resources get names of Kubernetes resources

ETCD backup and restore

  • endpoint?
  • when restore, change etcd files location in hostPath
    vi /etc/kubernetes/etcd.yaml

monitoring, logging

  • Metrics-server(CPU, memory), k top nodes, k top pods —containers=true
  • crictl commands, logs, containers
  • To load logs of service journalctl -u kubelet -f

DaemonSet vs deployment vs StatefulSet

  • DaemonSet 所有 Nodes 上运行 pod,每个 Node 上最多只能运行一个 pod.适合一些系统层面的应用，例如日志收集、资源监, eg: Kube-proxy
  • StatefulSet, StatefulSet 类似于 ReplicaSet，但是它可以处理 Pod 的启动顺序，为保留每个 Pod 的状态设置唯一标识

podAntiAffinity •

  • 仅在 worker 上运行 topologySpreadConstraints
  • requiredDuringSchedulingIgnoredDuringExecution,调度器只有在规则被满足的时候才能执行调度。此功能类似于 nodeSelector， 但其语法表达能力更强。
  • Node Affinity,

临时卷（Ephemeral Volume）

  • emptyDir, which is erased when a pod is removed,

Service CIDR range—> static-pod, kube-apiserver.yaml

  • cat /etc/kubernetes/manifests/kube-apiserver.yaml | grep range
  • /etc/kubernetes/manifests/kube-apiserver.yaml : –service-cluster-ip-range
  • /etc/kubernetes/manifests/kube-controller-manager.yaml : –service-cluster-ip-range

upgrade culster

kubeadm version
kubelet version
service kubelet status
systemctl daemon-reload && systemctl restart kubelet

drain node01, A forceful drain of the node will delete any pod that is not part of a replicaset.
kubectl cordon node01, This will ensure that no new pods are scheduled on this node and the existing pods will not be affected by this operation.

add node to cluster

kubeadm token create， node join

  1. controlplane node, kubeadm token create --print-join-command
  2. Worker node, execute the output from the above

Certificate

  • view cert details openssl x509 -noout -text -in ./server.crt
  • kubeadm command , kubeadm certs check-expiration
  • Renew apiserver cert , kubeadm certs renew api-server

Networking (CNI plugin)

  • default path /etc/cni/net.d/
  • Coredns-deployment, //in kube-system kubectl -n kube-system get pod
    Kube-scheduler, kube-scheduler.yaml

Kubeconfig

  • cluster and context info, ~/.kube/config

manually scheduling

  • If there is no scheduler in kube-system, add nodeName in pod.yaml, otherwise pod status is pending.

troubleshooting

  • sudo journalctl -u kubelet

Exam experience

5/2/2023, exam check in is more smooth than AWS and easy to use.

  1. Exam questions are way simpler than the killer mock exam. Only 1/3 amount of questions of the mock one.
  2. Only regret things is I use laptop for easy, it’s more hard for me do things quick and my eyes get blur. Next time, I will choose big monitor.
  3. Exam topics:
    • ETCD backup and restore
    • multiple containers pod
    • cluster upgrade
    • cluster troubleshooting
    • clusterrole, rolebinding and serviceaccount
    • PV, PVC, pod

6/2/2023

24 hours later, after I got the exam result, I realized that I failed the exam. I missed some small details when I am doing the exam. There are some knowledge gaps.