絶賛勉強中、適宜更新していきます。
作って捗ったエイリアス
alias kgp="kubectl get pod"
alias kc="kubectl"
alias kgs="kubectl get service"
alias kgd="kubectl get deploy"
alias kcd="kubectl delete"
alias kdd="kubectl delete deployment"
alias kds="kubectl delete service"
alias kdp="kubectl delete pod"
よく使うコマンド
create pod
kc run --image=<myimage> --restart=Never my-pod
create deployment
kc run --image=<myimage> --restart=Always --replicas=2 my-deployment
create service
kc expose deployment my-deployment --type=NodePort --port==3306
minikube のdocker daemon をローカルから操作
https://dzone.com/articles/running-local-docker-images-in-kubernetes-1
minikube ssh
# To point your shell to minikube's docker-daemon, run:
# eval $(minikube -p minikube docker-env)
to get out, just close the current shell, as
https://stackoverflow.com/questions/53010848/how-to-connect-to-local-docker-daemon
ローカルでdocker registory を動かしてそこにプッシュ
参考
docker run -d -p 5000:5000 \
-v ~/.dockerregistry:/var/lib/registry \
--restart always \
--name registry \
registry:2
docker tag progrhyme/compose2minikube:v1 localhost:5000/progrhyme/compose2minikube:v1
docker push localhost:5000/progrhyme/compose2minikube:v1
Node／Podのリソース使用状況確認
Nodeによる使用量確認： kubectl top node [Node名]
Podによる使用量確認： kubectl top pod [Pod名]
k8s Worker NodeまたはPodによる、ノードCPUとメモリの、実際の使用量・使用率を確認できる。
なお、kubectl describe ～コマンドでも、CPUとメモリのリソース状況は確認できる。
が、describeではk8sによって確保された値（余分含む）を出力するのに対し、
topでは実際の使用状況を出力する点で異なる。
＜オプション＞
Pod内の、コンテナごとのリソース使用量を確認したい時は、「--containers」オプションをつける。
kubectl top pod --containers
とりあえずpodをrunning でキープしたい
kc run --image=<my-image> --restart=Never input-client --command -- tail -f /dev/null
dry-run=client
kubectl run bee --image=nginx --restart=Never --dry-run -o yaml > bee.yaml
Imperative vs Declarative
important concepts.
get labels
kubectl get nodes node01 --show-labels
get tolerations
kubectl explain pod --recursice | grep -A5 tolerations
get taint
kubectl describe nodes master | grep -i taint
label nodes
kubectl label nodes node01 size=Large
scale deployments
kubectl scale deployment nginx --replicas=5
create deployment and scale
kubectl create deployment blue --image==nginx
kubectl scale deployment blue --replicas=6
get pods -o wide
kubectl get pods -o wide
get all daemon sets in all name spaces
kubectl get ds --all-namespaces
On how many nodes are the pods scheduled by the DaemonSet kube-proxy
kubectl -n kube-system get pods -o wide | grep proxy
describe daemonset info about image in kube-system
kubectl describe ds weave-net -n kube-system | grep -i image
deploy daemonset
first
kubectl create deployment <name> --dry-run=client -o yaml > elastic.yaml
then change
-
deploymenttoDaemonSet - remove
replica - remove
strategy - remove
status - set appropriate
namespace
static pod
static pod means in the node kubelet only manage pod without knowing about deployment, replica set like that.
kubelet only reads the manifest file from certain path.
get static pods
kubectl get pods --all-namespaces
and check suffix if it ends with namespace name.
get the path of static pod definition file
ps -ef | grep kubectl
and look for --config
then go to that path, and look for staticPodPath: by
grep -i static <path of --config>
create static pod named static-busybox and that uses busybox image and thte command sleep 1000
kubectl run --restart=Never --image=busybox static-busybox --dry-run=client -o yaml --command -- sleep 1000 > /etc/kubernetes/manifests/static-busybox.yaml
delete static pods created in some other node
first get ip for that node by
kubectl get nodes -o wide
second, ssh to the target node by
ssh <target-internal-ip>
follow
ps -ef | grep kubelet | grep "\--config"
and delete static yaml file.
deploy new scheduler to the cluster
cd /etc/kubernetes/manifests
cp kube-scheduler.yaml /root/my-scheduler.yaml
create new pod with custom scheduler
under the spec, add
schedulerName: my-scheduler
metric server getting started
minikube addons enable metrics-server
once it is configured,
kubectl top node
kubectl top pod
will show the memory consumption etc.
to keep watching,
watch "kubectl top node"
is handy.
see logs stream
kubectl logs -f <pod>
see logs with specific container
kubectl logs <pod> -c <container>
ENTRYPOINT in Dockerfile == command in k8s yaml
CMD in Dockerfile == args in k8s yaml
create yaml file based on existing pod
kubectl get pod <pod name> -o yaml > test.yaml
create configmap inperative way
kubectl create configmap <configmap name> --from-literal=key=value
useful for lookup
kubectl explain pods --recursive | grep envFrom -A3
create secret in imperative way
kubectl create secret <secret name> generic --from-literal=key=value
get pod and services
note **no space **
kubectl -n elastic-stack get pods,svc
take out the node for maintanance
kubectl drain node01 --ignore-daemonsets
drain basically means make node as unschedulable.
after the maintanance, bring back node
kubectl uncordon node01
after uncordon, only when new pods are scheduled, it may be assigned to this node.
kubectl version
kubectl version --short
upgrade kubeadm for the master
apt install kubeadm=1.18.0-00
kubeadm upgrade apply v1.18.0
apt install kubelet=1.18.0-00
"" upgrade kubeadm for the worker
apt install kubeadm=1.18.0-00
kubeadm upgrade node
apt install kubelet=1.18.0-00
Backup option 1: like Velero
kubectl get all --all-namespaces -o yaml > backup.yaml
Backup option 2: use etcd client etcdctl
default port for etcd is 2379
to see the help,
ETCDCTL_API=3 etcdctl snapshot save -h
then save it to
ETCDCTL_API=3 etcdctl --cacert="/etc/kubernetes/pki/etcd/ca.crt" --cert="/etc/kubernetes/pki/etcd/server.crt" --key="/etc/kubernetes/pki/etcd/server.key" --endp
oints=127.0.0.1:2379 snapshot save /tmp/snapshot-pre-boot.db
or first try
ETCDCTL_API=3 etcdctl member list --cacert="/etc/kubernetes/pki/etcd/ca.crt" --cert="/etc/kubernetes/pki/etcd/server.crt" --key="/etc/kubernetes/pki/etcd/server.key" --endpoints=127.0.0.1:2379
for restore from etcd
ETCDCTL_API=3 etcdctl snapshot restore -h
ETCDCTL_API=3 etcdctl --endpoints=https://[127.0.0.1]:2379 --cacert=/etc/kubernetes/pki/etcd/ca.crt \
--name=master \
--cert=/etc/kubernetes/pki/etcd/server.crt --key=/etc/kubernetes/pki/etcd/server.key \
--data-dir /var/lib/etcd-from-backup \
--initial-cluster=master=https://127.0.0.1:2380 \
--initial-cluster-token=etcd-cluster-1 \
--initial-advertise-peer-urls=https://127.0.0.1:2380 \
snapshot restore /tmp/snapshot-pre-boot.db
then
cd /etc/kubernetes/manifests/
vim etcd.yaml
- --data-dir=/var/lib/etcd-from-backup
- --initial-cluster-token=etcd-cluster-1
also change mountPath
node to node
see the ip address assigned to master node
ifconfig -a
cat /etc/network/interfaces
for MAC address,
ip link command to see the assigned hard ware address.
default route
ip route
to see the default gateway
port that scheduler is listening on
netstat -natulp | grep kube-scheduler
show how many ip can be assigned to node
ip addr show weave
when bash is not available when exec
kubectl exec -it <pod> -- sh
see default gateway of the pod
kubectl exec -it <pod> -- sh
then
ip r
ip address cidar range for nodes
ip a
and check cidar
ip address cidar range for pod
kubectl -n kube-system logs weabe-net-<> -c weave
IP address range for service
cat /etc/kubernetes/manifests/kube-apiserver.yaml
check roles and role bindings for service account
kubectl -n ingress-space get roles.rbac.authorization.k8s.io
kubectl -n ingress-space get rolebindings.rbac.authorization.k8s.io
see kubeadm token
kubeadm token create --help
kubeadm token create --print-join-command
around scheduler troubleshoot
cat /etc/systemd/system/kubelet.service.d/10-kubeadm.conf
see
KUBELET_CONFIG_ARGS=--config=/var/lib/kubelet/config.yaml
or
where the static pod yaml file located.
vim /etc/kubernetes/manifests/kube-scheduler.yaml
scale deployment
kubectl scale deployment app --replicas=2
scaleing, deployment is hadled by
kube-controller-manager-master
minikube dashboard
minikube addons enable dashboard
minikube dashboard --url
allow all the traffic to the proxy
kubectl proxy --address='0.0.0.0' --disable-filter=true
untaint master node
kubectl taint nodes myhost node-role.kubernetes.io/master:NoSchedule-