Overview
Stop chasing mysterious OOMKilled pods. In ten coffee-break lessons you’ll size memory the right way, read container cgroup stats, and react before your app is evicted.
First 3 emails are free; unlock the remaining 7 when you’re ready.
| Duration | Pace | Effort |
|---|---|---|
| 10 days | 1 email / day | ≈ 5 min / drill |
Who this is for
- Platform / DevOps engineers keeping clusters green
- Backend devs haunted by sporadic SIGKILLs
- SREs who need repeatable memory SLOs
Sneak-peek curriculum
| Day | Skill unlocked | 5-Minute Exercise |
|---|---|---|
| 1 | Spot an OOMKill fast | Deploy a pod with busybox → trigger head -c 150M /dev/zero → watch kubectl describe pod |
| 2 | Memory requests vs. limits | Rewrite the manifest: set request = 100 Mi, limit = 120 Mi, observe QoS shift |
| 3 | Node-level visibility | Install metrics-server (if missing) & run kubectl top pod,nodes for mem hot-spots |
| 4-10 | 🤫 Pro + Bonus drills | Cgroup-v2 stats • kubelet eviction thresholds • VerticalPodAutoscaler • cAdvisor / Prometheus alerts • oom_score_adj tuning • memory leak flamegraphs • node swap-free best-practice |
Full syllabus delivered after purchase — no guesswork.
Results you’ll ship
- Zero surprise OOMKills in production for memory-bound services
- Memory budgets that survive real traffic spikes
- Repeatable debug checklist your whole team uses
Get the full series — $9
Unlock days 4-10 now »14-day money-back guarantee. One-time payment, no subscription.
FAQ
Can I follow along without cluster admin rights?
Yes — all drills work on a local kind or minikube cluster, no cloud bill required.
Will these lessons change my live workloads?
Exercises run in isolated namespaces or local clusters; nothing touches prod unless you copy the manifests there.
How “advanced” is it?
If you can kubectl apply -f app.yaml, you’ll keep up fine. Each email links deeper docs for optional reading.
Still exploring? Browse all challenges