- Project State: Prototyping
- Issues Response SLA: None
- Pull Request Response SLA: None
For more information on project states and SLAs, see this documentation.
This plugin allows applications that rely on Train to communicate with the Kubernetes API. For example, InSpec uses this to perform compliance checks against Kubernetes Containers.
Train itself has no CLI, nor a sophisticated test harness. InSpec does have such facilities, so installing Train plugins will require an InSpec installation. You do not need to use or understand InSpec.
Train plugins may be developed without an InSpec installation.
Train plugins are installed using inspec plugin install
. Once released, train-k8s-container will be released as a RubyGem, like all train plugins.
While the project is still in its early phases and the gem is not yet released, you can preview the functionality.
First obtain a git clone of the train-k8s-container repo:
$ git clone https://github.com/inspec/train-k8s-container.git
Then use the path form of the plugin installer:
$ inspec plugin install path/to/train-k8s-container
train-k8s-container plugin installed via source path reference, resolved to entry point /Users/wolfe/sandbox/inspec/train-k8s-container/lib/train-k8s-container.rb
This technique allows you to run the plugin from the source in the given directory, so you can run from a branch or work from edits. If you are curious how this works, see ~/.inspec/plugins.json .
Once train-k8s-container is released as a gem, you can install it by name.
Simply run:
$ inspec plugin install train-k8s-container
Below are the two mandatory pre-requisites for this plugin to work.
train-k8s-container
creates connection to k8s containers using the kubeconfig- kubectl Both the kubeconfig and kubectl has to be present on the host machine from where this plugin is executed
By default, it looks for the kubeconfig file in ~/.kube/config
. It can be overridden by the ENV["KUBECONFIG"]
.
pod
and container_name
are two mandatory parameters. namespace
is optional only if the namespace is a default
k8s namespace.
You can then run the Inspec using the --target
/ -t
option, using the format -t k8s-container://<namespace>/<pod>/<container_name>
:
for example,
In order to connect to a container nginx
of pod shell-demo
running on prod
namespace
the URI would be k8s-container://prod/shell-demo/nginx
if in case the targeted container is on a default
namespace, it can be skipped like this k8s-container:///shell-demo/nginx
$ inspec detect -t k8s-container://default/shell-demo/nginx
────────────────────────────── Platform Details ──────────────────────────────
Name: k8s-container
Families: unix, os
Release: 1.2.1
Arch: unknown
$ inspec shell -t k8s-container:///shell-demo/nginx
Welcome to the interactive InSpec Shell
To find out how to use it, type: help
You are currently running on:
Name: k8s-container
Families: unix, os
Release: 1.2.1
Arch: unknown
inspec>
The intended usage of this plugin is to allow os
-platform-targeted profiles to run on Kubernetes containers. So, once you have connected, you should be able to run:
example usage of Inspec command resource
inspec> describe command("whoami") do
its("stdout") { should cmp "alice" }
end
example usage of Inspec file resource
inspec> describe file('/proc/version') do
its('content') { should cmp "Linux version 6.5.11-linuxkit (root@buildkitsandbox) (gcc (Alpine 12.2.1_git20220924-r10) 12.2.1 20220924, GNU ld (GNU Binutils) 2.40) #1 SMP PREEMPT Wed Dec 6 17:08:31 UTC 2023\n" }
end
Bugs, typos, limitations, and frustrations are welcome to be reported through the GitHub issues page for the train-k8s-container project.
You may also ask questions in the #inspec channel of the Chef Community Slack team. However, for an issue to get traction, please report it as a github issue.
If you wish to contribute to this plugin, please use the usual fork-branch-push-PR cycle. All functional changes need new tests, and bugfixes are expected to include a new test that demonstrates the bug.
Plugin Development is documented on the train
project on GitHub.