Skipper

Skipper is an interactive terminal tool for managing multiple kubernetes clusters.

Installation and Setup

Before you start, you will need to have python3 and pip3 installed. Then clone the repo, and run

chmod +x ./installer && ./installer

Running Skipper

In the project directory, run

./skipper

On startup, Skipper will locate your kube-config by looking at the KUBECONFIG environment variable. If this is empty, it will look at the default location (~/.kube/config).

Skipper loads your kube-config in order to get information from your clusters, and will never create, modify, or delete any of your kubernetes resources.

Logs

Logs are stored in the logs/ directory. flask.log contains output from the flask webserver, crawler.log contains output from the crawler, and skipper.log contains output from the frontend curses application.

Usage

Skipper usage walkthrough video: using skipper

navigation keybinds

-> or ENTER: navigate into a resource

<- or BACKSPACE: navigate out of a resource

[left pane] up and down arrows: go up / down a list of resources

[right pane] up and down arrows: scroll

cluster mode

Cluster mode presents your kubernetes resources in the following hierarchy:

cluster -> ns -> {deployment,daemonset,statefulset,service} -> pod

Note: currently standalone pods (those not being managed by a deployment) do not show up in cluster mode.

app mode

App mode presents your kubernetes resources in the following hierarchy:

Application -> Deployable -> {deployment,daemonset,statefulset,service} -> pod

Application and Deployable are custom resource definitions specified by IBM's Multicloud Manager. If none of your clusters have the Application or Deployable crds, app mode will be disabled for you.

anomaly mode

Anomaly mode presents a curated shortlist of pods across your clusters that are in an error state.

query mode

Query mode lets you search for resources across all your clusters. Currently query mode supports keyword search and filters. Supported filters are app, cluster, ns, kind. You can apply as many filters as you'd like in succession.

Example queries:

app:my-app

Will return all kubernetes resources that belong to the Application named my-app.

kind:pod frontend

Will return all pods with 'frontend' in the name.

cluster:iks-extremeblue ns:default

Will return all kubernetes resources that reside in the default namespace of the cluster named iks-extremeblue.

modeswitching

Whenever you have a resource selected and switch to app / cluster mode, Skipper will bring you to the location of the selected resource within the app / cluster hierarchy.

Architecture Overview

A high level overview of how Skipper works under the hood can be found here: skipper.pdf

Future Work

• Convert run script to a python script (currently in-progress on run-py branch)
• Package Skipper as a python package and upload to the Python Package Index
• Have Skipper automatically adjust when terminal window is resized
• Add additional search bar keybinds for highlighting, copy, paste, cut
• Add infrastructure mode to see the Node -> pod hierarchy
• Add ability to exec into containers within pods
• Add ability to open yamls / logs in preferred editor
• Script Skipper backend to allow users to run commands in terminal
• Add related resources to summary panel

Credits

Extreme Blue 2019, RTP Lab

Sponsored by Dave Lindquist and the IBM Cloud Private team

Interns

Name	School	Major	Role	Email
Patricia Lu	MIT	EECS	EB Technical Intern	pjlu01@gmail.com
Jane Hsieh	Oberlin College	CS (CSCI)	EB Technical Intern	jwhsieh98@gmail.com
Tom Gong	UT Austin	CS / Marketing	EB Technical Intern	tom.gong@utexas.edu
Caitlin Endyke	UMich	MBA / HCI	EB Offering Management Intern	endyke@umich.edu

Mentors

• Ross Grady (RTP EB Lab Manager)
• Shikha Srivastava
• Chris Waldon
• Ethan Swartzentruber
• Ryan Smith
• Yu Cao
• Jorge Padilla
• Kevin Myers
• Michael Elder
• Charles Quincy
• Sanjay Joshi