Kube DOOM: Kubernetes Chaos Engineering, 1993 Style#

What Fresh Hell Is This?#

Kube DOOM is a Kubernetes cluster management tool that transforms pod termination into a first-person shooter experience. Because apparently kubectl delete pod wasn't visceral enough.

Each pinky daemon in the game represents a pod in your cluster. When you frag a demon, the corresponding pod gets terminated. It's like chaos engineering, but with more pixelated blood and the iconic sound of a double-barreled shotgun.

Currently configured to target the whoami namespace, because apparently we need to answer life's fundamental question: "Who am I?" with "Someone who kills pods for fun."

Why Does This Exist?#

The Philosophical Answer#

Chaos engineering meets nostalgia. Resilience testing meets dopamine release. Your cluster meets the Doom Slayer.

The Practical Answer#

• Visual Pod Management: Sometimes you just want to see what you're killing
• Stress Relief: 3 AM incident? Kill some pods with style
• Team Building: Nothing says "DevOps culture" like a multiplayer deathmatch against your own infrastructure
• Training: Teach juniors about pod lifecycle management through the universal language of violence

The Honest Answer#

Because we can. Because it's hilarious. Because the cluster can handle it (hopefully).

Installation Status#

Kube DOOM is deployed to the kubedoom namespace with sync wave 86 (which, fun fact, is also the number of times you'll reload your shotgun before finding that one crashlooping pod).

Target Namespace: whoami

ArgoCD Application: 86-kube-doom.yaml

Getting Started: Your Journey Into Pod Purgatory#

Prerequisites#

• VNC client (TigerVNC recommended, RealVNC if you're fancy)
• A cluster with pods you don't mind killing
• A sense of humor
• Questionable decision-making skills

Step 1: Establish Port Forward#

First, tunnel into the hellscape:

1

kubectl port-forward -n kubedoom service/kubedoom 5900:5900

Step 2: Connect via VNC#

Launch your VNC client and connect to:

1

127.0.0.1:5900

VNC Password: idbehold

Step 3: Rip and Tear (Pods)#

1. Use arrow keys or WASD to navigate
2. Spacebar to open doors (because even in pod management, some things require manual intervention)
3. Ctrl/Left-click to fire your weapon of choice
4. Kill demons to terminate pods
5. Watch in horror/delight as pods respawn (if controlled by a Deployment)
6. Question your career choices
7. Realize this is the most fun you've had at work in months

Cheat Codes: Because Even Pod Management Needs Easy Mode#

The classic DOOM cheats work in Kube DOOM, because why not add invincibility to cluster administration?

Activation Instructions#

To enter cheat codes: 1. Simply type them during gameplay (no console needed) 2. Screen will flash briefly to confirm activation 3. Your conscience will flash warnings about production safety 4. Ignore both and proceed with extreme prejudice

Technical Details (The Boring But Necessary Part)#

How It Works#

1. Kube DOOM watches the target namespace (whoami) for pod events
2. Each pod is represented as a pinky daemon in the game
3. When you kill a monster, Kube DOOM executes kubectl delete pod
4. Pod controllers (Deployments, StatefulSets) will respawn pods - because true chaos never dies
5. You reload, respawn, and repeat

Architecture#

1
2
3
4
5
6
7

┌─────────────┐         ┌──────────────┐         ┌─────────────┐
│ VNC Client  │────────▶│  Kube DOOM   │────────▶│   Whoami    │
│  (You)      │  5900   │   Service    │  K8s API│  Namespace  │
└─────────────┘         └──────────────┘         └─────────────┘
      │                        │                         │
      │                        │                         │
      └──────[PEW PEW]───-─────┴────[kubectl delete]─────┘

Resource Configuration#

Kube DOOM is deployed via the upstream manifest from the official repository.

Namespace: kubedoom - quarantined for everyone's safety

Sync Wave: 86 - deployed late enough that you have pods to kill

Troubleshooting: When Things Go to Hell(spawn)#

Problem: I've Been Playing for 4 Hours#

Symptoms: Complete loss of time awareness, questioning life choices

Solutions: - This is working as intended - Take a break - Touch grass - Remember you have actual production issues to fix - Return in 10 minutes

FAQ: Frequently Asked Questionable Decisions#

Q: Is this safe for production?#

A: Define "safe." Define "production."

Q: Can I target multiple namespaces?#

A: Not in the current configuration. For multiple namespaces, you'd need to deploy multiple instances. Then you can have co-op multiplayer. Then you can explain to your manager why your monitoring shows "unusual pod churn."

Q: What happens if I kill a database pod?#

A: The same thing that happens when you kubectl delete a database pod, but with more explosions and the Doom Guy's grunting. StatefulSets will handle it. Probably. Maybe check your backups first.

Q: Can I stream this?#

A: Only if you're prepared to become either famous or unemployed. Possibly both.

Q: My boss saw me playing this. Help?#

A: "It's chaos engineering!" "We're testing pod resilience!" "This is active learning!" Or just update your resume. Your call.

Q: Is this better than kubectl delete pod?#

A: Objectively? No. Subjectively? Absolutely. Therapeutically? Without question.

The Philosophy of Kube DOOM#

In the end, Kube DOOM teaches us valuable lessons:

1. Pods are ephemeral - They live, they die, they respawn. Just like demons in Hell.
2. Chaos builds resilience - If your app can't survive random termination, it can't survive production.
3. DevOps should be fun - Who says cluster management has to be boring?
4. Visual feedback matters - Sometimes you need to see the destruction to feel alive.
5. Nostalgia is powerful - Combining 90s gaming with 2020s infrastructure is oddly satisfying.

Remember: In the grim darkness of the Kubernetes cluster, there is only DOOM.

Now if you'll excuse me, I have some demons to kill in the whoami namespace. They know who they are.