Here lies my swiss army knife of kubernetes tools I use on a daily-basis.
Please not all tools are built for BASH, SH and ZSH, but may be easy enough to port to other shells. Feel free to send PRs.
git clone https://github.com/caruccio/autokube
cd autokube- Install
- AutoKubectl: mnemonic kubectl command generator.
- AutoKubeconfig: switch KUBECONFIG file as you
cdinto a new directory. - ShowKubectl: print final kubectl command before executing it.
To install for the current user only (bash/sh/zsh) using the local repository:
make install-user
To install system-wide (bash/sh only):
sudo make install
Files will be installed into /etc/profile.d. Just start a new shell section to use it.
Files are sourceed from the current directory (where you cloned it) on your RC file (~/.bashrc, ~/.zshrc or ~/.profile).
Either start a new shell session or source it (ex: source ~/.bashrc or source ~/.zshrc)
Mnemonic kubectl command generator
Tired of typing kubectl get event -w -n=default -v=3 -o=json?
Try this instead: kgevwnvoj default 3
ALERT: This is not a shell alias
Inspired by ahmetb's kubectl-aliases, this tool will resolve an alias-like command to kubectl.
Take, for instance, both commands below.
kgevwnvoj default 3
kwojgevvn 3 defaultEach one will expand for the following commands respectively.
kubectl get event -w -n=default -v=3 -o=json
kubectl -w -o=json get event -v=3 -n=defaultWhere:
Mnemonic Resolves To Mnemonic Class
k --> kubectl
g --> get [verb]
ev --> event [resource]
w --> -w [option]
n --> namespace=$1 [option + $1]
v --> -v=$2 [option + $2]
oj --> -o=json [option]
The only rule is to follow the given order:
k -> [verb] -> [resource] -> [option|prefix|suffix]
There is no order within each class of mnemonics.
You can prefix and suffix to the final command. Mnemonic starting with - are inserted in front of the command, and those starting with + are appended.
kgpoojn-w1+gr kube-system apiserverBecomes:
watch -n 1 -- kubectl get pods -o=json --namespace="kube-system" | grep "apiserver"
+-----------+ +-+ +--+ +-----+ +-----------------------+ +----------------+
| | | | | |
[prefix + [N]] [verb] | [option] [options + $1] [append]
|
[resources]
Note
-wis treated special. It accepts a numeric value after thewto be used bywatch's parameter-n [N]. If[N]is omited, likekgpo-w, the default value2is used.
It defines the special bash function command_not_found_handle (command_not_found_handler for zsh) to intercept commands not found.
The function walks char-by-char of the first parameter (ths command itself), searching for the longest mnemonic sequence on a dictionary,
translating one or more characters into kubectl verbs, resources and options.
You can see all available mnemonics by executing kH (H stands for help).
The dictionaries can be expanded or modified with your custom mappings. It will look for configs in the fiels /etc/autokubectl and ~/.autokubectl
(or from file defined in env AUTOKUBECTLRC.
The file format is a simple yaml like this:
verb:
wapo: wait --for=condition=Ready pod/%s
poev: alpha events --for pod/%s
## https://kubepug.xyz/
pug: deprecations
resource:
hr: helmrelease
hrepo: helmrepository
options:
v5: -v=5
v6: -v=6
prefix:
dry: "echo DRY:"
suffix:
wcl: "--no-headers | wc -l"Now you can use your custom mnemonics:
$ kpug --> kubectl deprecations
$ kghran --> kubectl get hr --all-namespace
$ kgnsw-dry --> echo DRY: kubectl get namespaces --watch
$ kgpo+wcl --> kubectl get pods --no-headers | wc -lIf var AUTOKUBECTL_DRYRUN=true, then no command is executed, and the resulting expansion is shown in stdout:
$ AUTOKUBECTL_DRYRUN=true \
kgevwnvoj default 3
kubectl get event -w -n="default" -v="3" -o=jsonThis method may conflicts with other tools that install a "command not found" handler function.
For example, Fedora-like distros often comes with package PackageKit-command-not-found installed.
This packages provides the file /etc/profile.d/PackageKit.sh with a single function command_not_found_handle, which is called by bash when a command you typed is not found.
That is how it can suggest speelling corretions or even packages for that uninstalled command in your system.
You can see this sections of bash's manual for details.
The problem is that files from /etc/profile.d are sourceed on boot, with no garantees that our file will overwrite PackageKit-command-not-found's function.
That said, you can only have one "command not found" function handler (for now). Thus, either you uninstall PackageKit-command-not-found or source autokubectl.sh from your ~/.bashrc:
There are problems with this method? Of course there are!! Who do you think I'm?! Dennis Ritchie?!
Sometimes you will face ambiguity, but I can live with that... Fell free to fix and send me a PR.
For example kgno: is this kubectl get node or kubect get -n=$1 -o=$2?
Turns out the longest mnemonic matches first, thus no (2 chars) will match as node, not n (1 char) + o (1 char).
Use the command kH to show help.
$ kH
Usage: k[verb][resource][options...|-prefix...|+suffix...]
Verbs
-----
H: HELP
K: krew
Ki: krew install %s
a: apply
ar: api-resources
av: api-versions
c: create
ci: cluster-info
d: describe
dbg: debug -it %s
dbgno: debug -it --image=alpine "node/%s" -- chroot /host
doc: explain %s
docx: explore %s
drain: drain --delete-emptydir-data --ignore-daemonsets
ed: edit
evpo: alpha events --for pod/%s
ex: exec -i -t
g: get
gnoa: get node -L=kubernetes.io/arch,node.kubernetes.io/instance-type,topology.kubernetes.io/region,topology.kubernetes.io/zone,kubernetes.azure.com/agentpool,kubernetes.azure.com/storagetier,kubernetes.azure.com/storageprofile
gnoe: get node -L=kubernetes.io/arch,node.kubernetes.io/instance-type,topology.kubernetes.io/region,topology.kubernetes.io/zone,eks.amazonaws.com/capacityType
gnog: get node -L=kubernetes.io/arch,node.kubernetes.io/instance-type,topology.kubernetes.io/region,topology.kubernetes.io/zone,cloud.google.com/gke-nodepool,cloud.google.com/machine-family
gnok: get node -L=kubernetes.io/arch,node.kubernetes.io/instance-type,topology.kubernetes.io/region,topology.kubernetes.io/zone,eks.amazonaws.com/capacityType,karpenter.sh/capacity-type,karpenter.k8s.aws/instance-cpu,karpenter.k8s.aws/instance-memory
k: kustomize
lo: logs
lof: logs -f
lop: logs -f -p
p: proxy
pex: pexec -it -T %s
pexc: pexec -it -T %s -c %s
pexg: pexec -it -T %s --cnsenter-gc
pexgc: pexec -it -T %s -c %s --cnsenter-gc
pf: port-forward
pug: deprecations
rm: delete
run: run --rm --restart=Never --image-pull-policy=IfNotPresent -i -t --image=%s
sc: scale --replicas=%i
sh: exec -i -t %s -- sh -i -c "exec bash -i || exec sh -i"
shbr: exec -i -t %s -- apiclient exec -t control enter-admin-container
shc: exec -i -t %s -c %s -- sh -i -c "exec bash -i || exec sh -i"
t: top
tn: top node
tnp: top-node-pod
tp: top pod
ver: version
wapo: wait --for=condition=Ready pod/%s
z: fuzzy
Resources
---------
cm: configmap
cr: clusterrole.rbac.authorization.k8s.io
crb: clusterrolebinding.rbac.authorization.k8s.io
crd: crd.apiextensions.k8s.io
dc: deploymentconfig.apps.openshift.io
dep: deployment.apps
ds: daemonset.apps
ep: endpoints
ev: event
hr: helmrelease
hrepo: helmrepository
ic: ingressclass.networking.k8s.io
ing: ingress.networking.k8s.io
ingc: ingresscontroller.operator.openshift.io
is: imagestream.image.openshift.io
no: nodes
ns: namespaces
po: pod -n default
pr: prometheusrule.monitoring.coreos.com
prom: prometheus.monitoring.coreos.com
pv: pv
pvc: pvc
rb: rolebinding.rbac.authorization.k8s.io
ro: role.rbac.authorization.k8s.io
route: route.route.openshift.io
rs: replicaset.apps
sa: serviceaccount
sc: storageclass.storage.k8s.io
sec: secret
sm: servicemonitor.monitoring.coreos.com
sts: statefulset.apps
svc: service
Options
-------
L: -L=%s
all: --all
an: --all-namespaces
c: --container %s
dry: --dry-run=none -o=yaml
dryc: --dry-run=client -o=yaml
dryn: --dry-run=none -o=yaml
drys: --dry-run=server -o=yaml
f: -f=%s
force: --force
h: --help
i: -i
k: -k
l: -l=%s
n: --namespace=%s
nh: --no-headers
now: --now
o: -o=%s
oj: =@ojson
ojp: =@ojsonpath
ojs: =@ojson
ojson: -o=json
ojsonpath: -o=jsonpath=%s
on: =@oname
oname: -o=name
otemplate: -o=template=%s
otpl: =@otemplate
ow: =@owide
owide: -o=wide
oy: =@oyaml
oyaml: -o=yaml
oyml: =@oyaml
p: -p
r: --recursive
raw: --raw %s
rm: --rm
sb: --sort-by=%s
sl: --show-labels
sys: --namespace=kube-system
t: -t
v: -v=%s
v5: -v=5
v6: -v=6
w: --watch
Prefixes
--------
-: %s
-t: time
-w: watch -n %s --
Suffixes
--------
+: %s
+gr: | grep %s
Examples
--------
kgpo --> kubectl get pod -n default
kgpon kube-system --> kubectl get pod -n default --namespace=kube-system
kgpoansl --> kubectl get pod -n default --all-namespaces --show-labels
ksysgds --> kubectl --namespace=kube-system get daemonset.apps
kgposlan -v=6 --> kubectl get pod -n default --show-labels --all-namespaces -v=6
kafn pod.yaml default --> kubectl apply -f=pod.yaml --namespace=default
kgns-w3 --> watch -n 3 -- kubectl get namespaces
kgno- echo --> echo kubectl get nodes
kgno-+ "echo DRY: [" ] --> echo DRY: [ kubectl get nodes ]
Please refer to https://github.com/caruccio/autokube for instructions.Change current kubeconfig as you change directories.
$ cd /some/dir
Using kubeconfig: /some/dir/.kubeconfig
$ echo $KUBECONFIG
/some/dir/.kubeconfigYou can also use kubecfg directly to update $KUBECONFIG env var:
$ kubecfg ./kubeconfig-dev
Using kubeconfig: /some/dir/.kubeconfig-devRunning kubecfg alone prints the current $KUBECONFIG:
$ kubecfg
/some/dir/.kubeconfig-devThe flag -u unsets $KUBECONFIG:
$ kubecfg -u
Unsetting KUBECONFIG=/some/dir/.kubeconfig-devPrint full kubectl command before execute it.
It's just an small function which replaces kubectl and print the whole command to stderr.
Now every kubectl command, including aliases and functions, will be printed to stderr before it's executed.
$ kgno
+ kubectl get nodes