kubemrr mirrors description of Kubernetes resources for blazingly fast auto-completion.
Feel free to suggest improvements, I'm here to work on them 👍
Bash has the programmable completion feature which permits typing a partial command,
then pressing the [TAB] key to auto-complete the command sequence.To use the feature, you need to write a completion script
and place it in /etc/bash_completion.d folder (docs,
tutorial).
Fortunately, kubectl comes with a command to generate completion script (kubectl completion).
The script works extremely well, but slow, because each time you hit [TAB] it sends a request
to Kubernetes API Server. If your server is on a different continent you might wait for up to 2 seconds.
To reduce the delay, kubemrr keeps names of resources locally. See how to make a completion script which talks to kubemrr
instead of real Kubernetes API server in the example below.
To start watching servers, give context names from your kubeconfig file:
kubemrr watch dev prod
To make completion script that talks to kubemrr shell:
alias kus='kubectl --context us'
kubemrr completion bash --kubectl-alias=kus > kus
sudo cp kus /etc/bash_completion.d
Note that you need to have bash completion installed. It shoud be available on a Linux distribution. On a Mac,
install with brew install bash-completion.
Replace bash with zsh in the above command to generate completion script for zsh shell.
To test it:
source kus
kus get po [TAB][TAB]
kus get svc [TAB][TAB]
kus get deployments [TAB][TAB]
kus get configmaps [TAB][TAB]
kus get namespaces [TAB][TAB]
kus get nodes [TAB][TAB]
To make completion script that talks to kubemrr that is running on different host (use IP to save time on name resolution):
kubemrr completion bash --address=10.5.1.6 --kubectl-alias=kus > kus
- OSX:
curl -O https://raw.githubusercontent.com/mkokho/kubemrr/v1.3.0/releases/darwin/amd64/kubemrr
- Linux:
curl -O https://raw.githubusercontent.com/mkokho/kubemrr/v1.3.0/releases/linux/amd64/kubemrr