Before we launch kubed-sh, let's get some examples set up:

git clone https://github.com/mhausenblas/kubed-sh.git && cd kubed-sh/tc/node/

Next, to launch kubed-sh, simply execute:

kubed-sh

Now you should see kubed-sh pre-pulling container images and then you're good to go. This process takes some 30 seconds to complete and is only required the first time you launch kubed-sh.

Start by entering help and then try launching a script like so:

node test.js &

See the status:

ps

Invoke it:

curl test

You can leave kubed-sh at any point in time using exit. See also the usage reference for more details what you can do.

That's it, in a nutshell. With kubed-sh, you can use a Kubernetes cluster just like your local machine, without learning new concepts or commands.

Here are a few areas where kubed-sh does go beyond the local shell analogy:

• In itself kubed-sh is stateless. In other words, all the state is kept in Kubernetes via labels, see also the design.
• It extends the concept of an environment to something you can create and actively switch to.
• For long-running processes like an app server you can use the hot-reload feature (since 0.4) that makes kubed-sh watch for chances, allowing you to edit your sources locally and update them immediately on save into the Kubernetes cluster. See also the link to the demo below.

Learn more about how to use kubed-sh here:

• Introducing kubed-sh (5min video on YouTube)
• kubed-sh demo: hot-reload (2min video on Loom)