Ktools

Ktools is a set of containerized kubernetes-related tools and convenience functions. It is an easier way to manage your tools across dev, ci builds, deploys, and ops as well as across teams and platforms. You can also switch versions of tools on the fly.

Prereqs

All you need is bash, some version of docker (the client and the daemon), grep and sed.

Install Ktools

You can install ktools via the ktools docker image.

docker run premiereglobal/ktools:latest wrapper > ktools.sh && chmod u+x ktools.sh

Tools it provides

• kops - for managing the k8s cluster itself
• kubectl - for managing workloads on the cluster
• helm - for installing a set of k8s resources
• stim - for interacting with vault to get kube config or login to aws or get secrets from vault before deploying
• vault - for secret management
• terraform - for managing cloud infrastructure as code

Usage

Either download the ktools.sh script and make it executable or clone the ktools repo.

Then you can execute commands via ktools

cd path/to/ktools/
./ktools.sh kubectl config current-context
./ktools.sh helm ls

Or you can source ktools which allows you to run commands like you would with normally installed tools. Sourcing ktools effectively installs the tools by adding functions to your shell which run the tools in a docker container. Technically it lazy installs a tool the first time you run it.

cd path/to/ktools/
. ./ktools.sh
kubectl config current-context
helm ls

The best way to use ktools is to source it in your .bashrc or .bash_profile and have it indicate the current k8s cluster you are pointed at.

echo ". ~/path/to/ktools/ktools.sh --ktools-modify-prompt" > ~/.bashrc

kvm

If you have sourced ktools, it can act as a version manager for the tools it provides. If you aren't able to switch to a version, check that the version is valid and also check that the docker image for the version has been published. See the links at the top of ktools to the DockerHub pages for each tool

kversion kubectl # print the current version of the kubectl client
kvm kubectl 1.16.14 # switch to kubectl 1.16.14

kversion ks8 # print the version of k8s on the current cluster you are pointed at

kvm let's you specify a version of 'auto' for kubectl and vault (and helm if using the older helm 2). It will hit the currently configured server/cluster to determine the corresponding version of the client that is needed.

kvm kubectl auto # use the same kubectl version as the cluster's k8s version
kvm vault auto # use the vault cli version that matches the vault server

Updating Ktools

The ktools update command will run the latest docker image with the args to print out the ktools script and write over the current file in place.

# via executing ktools
./ktools.sh update

# after sourcing ktools
ktools update

Release A New Version

The ktools docker image is used to install or update ktools. It has commands for printing out the ktools version or the whole ktools script. The update command checks if the local ktools version is different than the version in the latest docker image.

Bump The Version

After making changes to ktools, make sure to bump the KTOOLS_VERSION env variable at the top of ktools.sh

Build The Image

docker build -t premiereglobal/ktools:latest .

Push

# if not logged in with an account that has permission to push to the premiereglobal org
docker login

docker push premiereglobal/ktools:latest

Gotchas

There are a few limitations when running cli tools in a docker container.

The main limitation is that the current working directory is mounted into the container which means you can't reference stuff outside the current working dir. The relevant dot files and dirs to the tool are also bind mounted. For example if you are using ktools and try kubectl apply -f ../mytemplate.yaml, it is not going to work because mytemplate.yaml is not in the directory that was mounted in the container. Ktools could technically be updated to handle this case but it currently does not handle it correctly.

You also can't bind mount when already inside a docker container. So if you are running a script inside a docker container that uses ktools, you can't use the containerized version of the tools.

Docker For Desktop on Mac (and maybe Windows) uses a VM that is detached from the host machine network so features like kubectl port-forward or proxy that rely on a port being exposed require native kubectl

Initial Authors/Contributors

• Devin Wilson (dwilson6)
• John Langewisch (jahndis)
• Craig Lewis (clewis)
• Nico Pampe (NicoPampe)
• Justin Martinez (jmartine)
• Kenji Chapa (kchapa)