Skip to content

Latest commit

 

History

History
281 lines (195 loc) · 11.7 KB

README.md

File metadata and controls

281 lines (195 loc) · 11.7 KB
Raw

Kubernetes tips and tricks

Kubernetes new releases

https://sysdig.com/?s=What%E2%80%99s+new+in+Kubernetes++

kubectl commands

Official kubectl Cheat Sheet

https://kubernetes.io/docs/reference/kubectl/cheatsheet/

View resource usage

kubectl top no
kubectl top po

Display docker image tage and SHA

kubectl get pod <my-pod-name> -ojson | jq '.status.containerStatuses[] | { "image": .image, "imageID": .imageID }'

Display logs for previous started container to debug abnormal successive restarts

kubectl logs <my-pod> --previous

Display http requests made by kubectl to kube-api

kubectl get po -v=6

Extract a token from a secret

kubectl get secret <my-secret> -n<my-namespace> -ojsonpath='{.data.token}' | base64 -d

Extract the content of a file from a secret (dot in file name must be escaped by \)

kubectl get secret <my-secret> -ojsonpath='{.data.jmxremote\\.password}' | base64 -d

Copy an object from one namespace to another

kubectl get secrets <my-secret> -ojson -n<my-src-namespace> | jq '.metadata.namespace = "<my-dest-namespace>"' | kubectl create -f -

Clean up an helm release manually

kubectl get deploy,sts,cm,secret,pvc,svc -oname -lrelease=<my-helm-release> | while read name; do kubectl delete $name; done

Wait for pod to be ready

kubectl wait po <my-po> --for=condition=Ready

Watch pods

watch kubectl get po -lrelease=<my-helm-release>

Find pods by date with jq

DEPLOYMENT_STARTDATE=`jq -n 'now'`
kubectl get po -lrelease=<my-helm-release> -ojson | jq -r --arg deployment_startdate $DEPLOYMENT_STARTDATE '.items[] | select(.metadata.creationTimestamp | fromdate | tostring > $deployment_startdate) | .metadata.name'

Find pods using a specific environment variable in secret

kubectl get po -ojson | jq -r '.items[] | select(.spec.containers[].env[]?.valueFrom.secretKeyRef.key=="<MY_VAR_ENV_NAME>") | .metadata.name'

Find deployments using a specific environment variable in secret

kubectl get deploy -ojson | jq -r '.items[] | select(.spec.template.spec.containers[].env[]?.valueFrom.secretKeyRef.key=="<MY_VAR_ENV_NAME>") | .metadata.name'

Inject an environment variable in a deployment

kubectl set env deployment/registry STORAGE_DIR=/local

Restart pods properly with a rollout (from 1.15)

kubectl rollout restart deploy <my-deploy>

Check if I'm allowed to do an action

kubectl auth can-i exec pod

Suspend all cronjobs at once

kubectl get cj -oname | while read name; do kubectl $name -p '{"spec":{"suspend":true}}'; done

Restart pod launched by job

kubectl get job <my-job> -o json | jq 'del(.spec.selector)' | jq 'del(.spec.template.metadata.labels)' | kubectl replace --force -f -

Patch all Terminating PVC to remove the finalizer

NS=mynamespace
kubectl -n $NS patch pvc $(kubectl -n $NS get pvc --no-headers | grep Terminating | awk '{print $1}') -p '{"metadata":{"finalizers":null}}'

List image in a deployment

kubectl get deploy -lrelease=si-labo -ojson | jq .items[].spec.template.spec.containers[0].image

List all image references in a namespace

kubectl get deploy -ojson | jq -r '.. | .image? // empty' | sort -u

List pods in status other than Running or Completed

kubectl get po -owide -A | grep -v 'Running\|Completed'

List evicted pods on all cluster

  • kubectl get po --field-selector=status.phase=Failed -A -owide
  • kubectl get po -A -ojson | jq -r '.items[] | select(.status.reason=="Evicted") | .metadata.namespace + " " + .spec.nodeName + " " + (.spec.priority|tostring)+ " " + .metadata.name + " : " + .status.message' | sort -k2,2 -k3nr

List pods with anti affinity

kubectl get po -ojson | jq '.items[] | select(.spec.affinity.podAntiAffinity!=null) | .metadata.name'

List pods with a guaranteed qos

kubectl get po -ojson | jq '.items[] | select(.status.qosClass=="Guaranteed") | .metadata.name'

List prority classes sort by value

kubectl get pc -ojson | jq -r '.items[] | .metadata.name + " : " + (.value|tostring)' | sort -k3nr

List priority infos for all pods sort by value

kubectl get po -ojson | jq -r '.items[] | .metadata.namespace + " : " + .spec.nodeName + " : " + .metadata.name + " : " + .spec.priorityClassName+ " : " + (.spec.priority|tostring)' | sort -k9nr -k5

List pods by restart count

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

List pods by age

kubectl get po --sort-by=.status.startTime

List all OOMKilled pods

kubectl get po -A -ojson | jq -r '.items[] | select(.status.containerStatuses[0].lastState.terminated.reason=="OOMKilled") | .metadata.namespace + " " + (.status.containerStatuses[0].restartCount|tostring) + " " + .metadata.name' | sort -k1,1r -k2nr

List all pods with privileged mode

kubectl get po -ojson -A | jq '.items[] | select(.spec.containers[].securityContext.privileged==true) | .metadata.namespace + " : " + .metadata.name'

List all pods using host's IPC namespace

kubectl get po -ojson -A | jq '.items[] | select(.spec.hostIPC==true) | .metadata.namespace + " : " + .metadata.name'

List all pods using host's network namespace

kubectl get po -ojson -A | jq '.items[] | select(.spec.hostNetwork==true) | .metadata.namespace + " : " + .metadata.name'

List all pods using host's PID namespace

kubectl get po -ojson -A | jq '.items[] | select(.spec.hostPID==true) | .metadata.namespace + " : " + .metadata.name'

Test anonymous access to health endpoints

KUBE_API_URL=<KUBE API URL>
for ep in version healthz livez readyz; do curl -k $KUBE_API_URL/$ep?verbose; done

List anonymous access

kubectl get clusterrolebindings -o json | jq '.items[] | select(.subjects? // [] | any(.kind == "User" and .name == "system:anonymous" or .kind == "Group" and .name == "system:unauthenticated"))'

Force delete a pod stuck in terminating status

kubectl delete pod <pod> --grace-period=0 --force

Force delete all pods stuck in terminating status at once

kubectl get po -owide | grep 'Terminating' | awk -F ' ' '{print $1}' | while read name; do kubectl delete po $name --grace-period=0 --force; done

List pre hook jobs for a release

kubectl get jobs -ojson | jq -r '.items[] | select(.metadata.annotations["helm.sh/hook"] and (.metadata.annotations["helm.sh/hook"]|contains("pre")) and .metadata.labels.release=="<my-helm-release>") | .metadata.name'

List pod owned by a hook job

kubectl get po -ojson | jq -r '.items[] | select(.metadata.ownerReferences[].name == "<my-hook-name>") | .metadata.name'

Delete succeeded jobs

kubectl get jobs -ojson | jq -r '.items[] | select(.metadata.annotations["helm.sh/hook"] and .status.succeeded==1) | .metadata.name' | while read name; do kubectl delete jobs $name ; done

List nodes with memory or disk pressure Taint Based Evictions

kubectl get no -ojson | jq -r '.items[] | select(.spec.taints!=null and (.spec.taints[0].key|contains("pressure"))) | .metadata.name + " : " + .spec.taints[0].key'

List allocated ressources per node

kubectl get no --no-headers | awk '{print $1}' | xargs -I {} sh -c 'echo {}; kubectl describe node {} | grep Allocated -A 5 | grep -ve Event -ve Allocated -ve percent -ve -- ; echo' for i in {01..12}; do echo dbk-k8s-worker-dev-${i}v; kubectl describe node dbk-k8s-worker-dev-${i}v|grep -A6 'Allocated resources:'; done

Drain a node

kubectl drain <node> --ignore-daemonsets --force --delete-local-data

ctr commands

List images

ctr --namespace k8s.io i ls

Push an image to a private registry

ctr --namespace k8s.io i push -u <user>:<password> <image>

Official helm Cheat Sheet

https://helm.sh/docs/intro/cheatsheet/

helm commands

Render chart templates locally

helm template . --output-dir=output-dir

List resources of a release

helm status <release> --show-resources

Get all values of a release

helm get values <release> -a

Get manifests of a release

helm get manifest <release>

Official flux Commands list

https://fluxcd.io/flux/cmd/

flux commands

Display all flux objects to check cluster health

flux get all -A

Display all not ready flux objects

flux get all -n <namespace> --status-selector="ready=false"

Display objects managed by a kustomization

flux tree ks <kustomization> -n<namespace>

Force kustomization reconciliation

flux reconcile ks <kustomization> -n<namespace> --with-source

Force helm release reconciliation

flux reconcile hr <release> -n<namespace> --with-source

Display diff between local and server kustomization

flux diff ks <kustomization> --path=<local path to kustomization> -n<namespace>

Search for drifts in reconciliation (helm revisions should increase periodically in case of an unwanted drift)

flux get hr -ndev --no-header | awk '{print $1}' | while read name; do helm history $name --max 1; done

Search for replicas specified in helm manifests (no defined replicas allow to scale without any detected drift)

flux get hr -ndev --no-header | awk '{print $1}' | while read name; do bash -c "echo $name && helm get manifest $name | grep replica"; done

Some recipes

Browse google registry

https://console.cloud.google.com/gcr/images/google-containers/GLOBAL

Ease kubectl use

Test

Add kubectl aliases

https://github.com/ahmetb/kubectl-aliases

Decode all secret content easily with ksd

https://github.com/ashleyschuett/kubernetes-secret-decode

kubectl get secret my-secret -o yaml | ksd

Add fuzzy search to your command with fzf

https://github.com/junegunn/fzf

kubectl get po | fzf

Read logs from all replicas at a time with stern

https://github.com/wercker/stern

Advanced Kubernetes Objects You Need to Know

https://engineering.opsgenie.com/advanced-kubernetes-objects-53f5e9bc0c28

Interact with kube-api like any other API

https://thenewstack.io/taking-kubernetes-api-spin/

How to terminate a side-car container in Kubernetes Job

https://medium.com/@cotton_ori/how-to-terminate-a-side-car-container-in-kubernetes-job-2468f435ca99

Docker Awareness in Java

Docker-in-Docker on Kubernetes

https://applatix.com/case-docker-docker-kubernetes-part-2/

How To Back Up and Restore a Kubernetes Cluster using Heptio Ark

https://www.digitalocean.com/community/tutorials/how-to-back-up-and-restore-a-kubernetes-cluster-on-digitalocean-using-heptio-ark

Configuring the Kubernetes CLI by using service account tokens

https://www.ibm.com/developerworks/community/blogs/fe25b4ef-ea6a-4d86-a629-6f87ccf4649e/entry/Configuring_the_Kubernetes_CLI_by_using_service_account_tokens1?lang=en

Treat your pods according to their needs - three QoS classes in Kubernetes

https://cloudowski.com/articles/three-qos-classes-in-kubernetes/

Prometheus Operator

https://github.com/helm/charts/tree/master/stable/prometheus-operator https://github.com/coreos/prometheus-operator/tree/master/Documentation https://sysdig.com/blog/kubernetes-monitoring-prometheus-operator-part3/

Kube eagle: prometheus exporter and grafana dashboard for a nice overview of a cluster

https://github.com/cloudworkz/kube-eagle

Checklist of production ready best practices for a kubernetes cluster

https://learnk8s.io/production-best-practices

Lens a great Kubernetes IDE

https://k8slens.dev/

Resources