Hackademy

Kubernetes Simulator missions

Kubernetes Simulator: 54 missions across 9 apps, from beginner to advanced, plus a sandbox to try anything. Free in your browser, with nothing to install.

Each mission drops you into a broken or unfinished system with a ticket to close. It checks the result, not your keystrokes, so any fix that works passes.

Open the Kubernetes Simulator

Static website

beginner, 7 missions

Put nginx on the cluster, give it an address, scale it for launch day and ship v2 — then survive the 3am page.

  1. Your first podbeginner, about 5 min
  2. Pods that healbeginner, about 8 min
  3. Give it an addressbeginner, about 8 min
  4. Look insidebeginner, about 8 min
  5. Launch daybeginner, about 8 min
  6. Ship v2intermediate, about 10 min
  7. The 3am pageintermediate, about 12 min

REST API

intermediate, 7 missions

Config and secrets, a crash loop, health checks, memory limits, a launch-day traffic spike and a rotated password — for a service people depend on.

  1. Config, not codeintermediate, about 10 min
  2. Read at startintermediate, about 10 min
  3. CrashLoopBackOffintermediate, about 10 min
  4. Ready when it's readyintermediate, about 10 min
  5. Out of memoryintermediate, about 10 min
  6. Launch-day trafficadvanced, about 12 min
  7. Rotate the passwordadvanced, about 12 min

Batch pipeline

intermediate, 6 missions

Jobs that run to the end and stop: retries, parallel work, a nightly CronJob, runs that pile up, and a 2am failure to backfill.

  1. Run it onceintermediate, about 8 min
  2. When a job failsintermediate, about 10 min
  3. Many handsintermediate, about 10 min
  4. Every nightintermediate, about 10 min
  5. Pile-upadvanced, about 10 min
  6. The 2am failureadvanced, about 12 min

Multiplayer game server

advanced, 7 missions

Servers with names that stay put, saves that survive a restart, a one-server canary, draining a node without emptying the arena, a server stranded by its own disk — and volumes kept past their claims and bound back to them.

  1. Every server has a nameadvanced, about 10 min
  2. Saves that surviveadvanced, about 12 min
  3. One server firstadvanced, about 10 min
  4. Don't empty the arenaadvanced, about 12 min
  5. Strandedadvanced, about 10 min
  6. Saves that surviveadvanced, about 10 min
  7. Claim it backadvanced, about 15 min

Microservices shop

intermediate, 7 missions

A frontend and two APIs that have to find each other: Services and cluster DNS, the port a Service forwards to, one Ingress in front of it all, NetworkPolicies that decide who may talk to whom and what may leave, a NodePort, and an ExternalName for a database outside the cluster.

  1. Find each otherintermediate, about 8 min
  2. Wrong doorintermediate, about 8 min
  3. One front doorintermediate, about 10 min
  4. Need to knowadvanced, about 12 min
  5. Outbound onlyadvanced, about 15 min
  6. A fixed addressintermediate, about 10 min
  7. Moved outintermediate, about 10 min

Multi-team platform

advanced, 5 missions

One cluster, two teams: namespaces to keep them apart, RBAC for a CI robot and an on-call engineer, and quotas so neither can starve the other.

  1. A room eachintermediate, about 8 min
  2. A key for the robotintermediate, about 10 min
  3. Look, don't touchintermediate, about 8 min
  4. Fair shareadvanced, about 12 min
  5. No room to rolladvanced, about 12 min

Node fleet

advanced, 8 missions

Which pod runs where: labels and selectors for the right hardware, taints to reserve a node, requests that must fit, a node emptied for a kernel patch — then replicas kept apart, spread evenly, steered by node affinity, and priority deciding who gets evicted.

  1. The right hardwareintermediate, about 8 min
  2. Reserved seatingintermediate, about 10 min
  3. It doesn't fitintermediate, about 8 min
  4. Kernel patchadvanced, about 12 min
  5. One basketadvanced, about 15 min
  6. All on oneadvanced, about 15 min
  7. Not on spotadvanced, about 12 min
  8. Checkout firstadvanced, about 15 min

Payments service

advanced, 4 missions

A JVM service that moves money: run its migration before the release, give a slow starter time to boot, look inside a pod with no shell, and pull back a bad release mid-rollout.

  1. Migrate, then releaseintermediate, about 10 min
  2. Not dead, just slowadvanced, about 12 min
  3. No shelladvanced, about 15 min
  4. Charged twiceadvanced, about 15 min

Projects

advanced, 3 missions

Whole launches from an empty cluster, written as tickets: a web store with an API and launch-day traffic, a multiplayer game with stateful servers, and an AI model on GPUs. No starter files, no commands in the hints.

  1. Launch the bookstoreadvanced, about 30 min
  2. Launch the arenaadvanced, about 30 min
  3. Serve the modeladvanced, about 30 min

AboutGuidesCompare

PrivacyCookiesTerms

Hackademy is a trading name of Mather Software Ltd, registered in England and Wales, company number 17443861. ICO registration ZC256628.

Contact: support@mathersoftware.com© 2026 Mather Software Ltd