Daniele Polencic
banner
danielepolencic.com
Daniele Polencic
@danielepolencic.com
Teaching Kubernetes at @LearnKube.com
Just landed: Learn Kubernetes weekly 170! My top picks:

💾 Persistent Storage in AKS
🌐 Kubernetes Gateway API Setup
🔗 Unifying Inner & Outer Loops
🔍 Korrel8r Observability Correlation Engine

Read it here: https://kube.today/issues/170
February 11, 2026 at 12:11 PM
Last year, I mapped the K8s multitenancy spectrum. This year, I'm adding AI agent sandboxing — where does it fit when the tenant isn't a human but a machine acting autonomously?

In collaboration with @vcluster.com

Feb 24 — 10:00 AM ET
https://ku.bz/Kzd7BPBqq
February 10, 2026 at 3:21 PM
When the cluster is full, priority decides who stays

Higher-priority pods can preempt lower-priority ones. The evicted pods go back to Pending

Full guide with 30+ diagrams: https://learnkube.com/kubernetes-scheduler-explained

Weekly newsletter: https://kube.today/learn-kubernetes-weekly
February 9, 2026 at 1:46 PM
Here's what most people miss: resource requests

Without them, the scheduler treats your pod as having zero size. Every node looks like it has infinite capacity

It will pack pods until the kubelet starts evicting them
February 9, 2026 at 1:46 PM
Scoring ranks the survivors from 0 to 100 across 13 plugins

A node that's mostly empty scores higher than one that's nearly full

Highest total wins. The pod gets scheduled there
February 9, 2026 at 1:46 PM
Example: your pod needs a GPU

The filter phase discards every node without one

Only GPU-equipped nodes survive to the scoring round
February 9, 2026 at 1:46 PM
Filtering is pass/fail

Wrong labels? Out. Not enough CPU? Out. Tainted? Out

A node must pass all 13 filter plugins to stay in the game. One failure and it's eliminated
February 9, 2026 at 1:46 PM
The scheduler picks pods from the queue one by one

For each pod, it runs two phases:

➀ Filtering — discard nodes that can't run the pod
➁ Scoring — rank what's left

Then it binds the pod to the winning node
February 9, 2026 at 1:46 PM
When you deploy a workload, Kubernetes needs to decide which node runs it

Pods start as "Pending" — they exist in etcd but aren't assigned anywhere yet

The scheduler watches for these unassigned pods and queues them
February 9, 2026 at 1:46 PM
The Kubernetes scheduler doesn't randomly assign pods to nodes

It runs a two-phase pipeline: filter, then score

Here's how it works, in 7 diagrams
February 9, 2026 at 1:46 PM
Just landed: Learn Kubernetes weekly 169! My top picks:

🔧 Vanilla K8s on NixOS
📊 Signals to Reliability: SLOs, Runbooks and Post-Mortems
🚪 Introducing MCP-Gateway in Kagenti
🤖 Kubently: Agentic K8s Debugging

Read it here: https://kube.today/issues/169
February 4, 2026 at 12:11 PM
Just landed: Learn Kubernetes weekly 168! My top picks:

🔍 Job Failures to Serial Console Bottlenecks
📊 OpenTelemetry-Native LGTM Stack
📦 Container Image Reduction: 1.2GB to 100MB
🔎 Lynq: SQL-like Kubernetes

Read it here: https://kube.today/issues/168
January 28, 2026 at 12:11 PM
💰 A full Kubernetes cluster for $2/month. Three nodes, multi-region, real workloads

💡 The clever part: skip the $10/month load balancer entirely. Tailscale operator + tailnet = fully private cluster by default

https://ku.bz/HpVyQMVv0
January 27, 2026 at 12:56 PM
Just landed: Learn Kubernetes weekly 167! My top picks:

🏗️ 80M User Infrastructure with a 5-Person Team
🤖 Self-Hosted RAG on Kubernetes
☁️ Managed Kubernetes on Proxmox with CAPI
📊 Prom Analytics Proxy

Read it here: https://kube.today/issues/167
January 21, 2026 at 12:11 PM
Just landed: Learn Kubernetes weekly 166! My top picks:

🐛 Fixing Kubelet Sandbox Errors
📊 Distributed Tracing with Tempo on AKS
💰 SimKube for Cost Forecasting
🚀 Kargo: GitOps Automation

Read it here: https://kube.today/issues/166
January 14, 2026 at 12:11 PM
83% of Kubernetes teams use network policies. 60% struggle to understand what they're actually doing

I surveyed 530 practitioners and here is what I found: https://kube.today/kubernetes-network-policies-2025
January 12, 2026 at 12:46 PM
Just landed: Learn Kubernetes weekly 165! My top picks:

🌐 Stretching a Layer 2 network over KubeVirt
💰 Reducing Feature Environment Costs with KEDA
☁️ Build your own SaaS with Kamaji
🔄 VolSync PV Replication

Read it here: https://kube.today/issues/165
January 7, 2026 at 11:36 AM
Just landed: Learn Kubernetes weekly 164! My top picks:

🛡️ Enforcing policies with GateKeeper
📦 Partitioned Log Storage on S3
🎮 Dynamic MIG Partitioning
🔐 Crowdsec: security solution

Read it here: https://kube.today/issues/164
December 31, 2025 at 11:36 AM
Just landed: Learn Kubernetes weekly 163! My top picks:

💾 Home Data Lake on a Budget
🧠 Smart Load Balancing at Databricks
🔗 Postgres Pooling with pgBouncer
🚪 Goma Gateway: Declarative API Gateway

Read it here: https://kube.today/issues/163
December 24, 2025 at 11:36 AM
Just landed: Learn Kubernetes weekly 162! My top picks:

🔐 Securing K8s with Kong + Keycloak
🌐 Networking for EKS Hybrid Nodes
📊 Victoria Logs vs Loki: Benchmarking Results
🛡️ Kexa: Cloud Compliance at Scale

Read it here: https://kube.today/issues/162
December 17, 2025 at 11:36 AM
Just landed: Learn Kubernetes weekly 161! My top picks:

🧩 JVM Crash-Loop Debugging
🏪 Edge Kubernetes Evolution
☁️ Kubernetes for Cloud Infrastructure
🏷️ Kube-vip Load Balancer

Read it here: https://kube.today/issues/161
December 10, 2025 at 11:56 AM
There's so much more that didn't make it to this list!

I'll take a short break to recharge and be ready for 2026!

Thank you all for an incredible year 🎉
December 8, 2025 at 2:06 PM
LLMs changed how I work this year

I can be way more ambitious with my projects now

I explored this in a webinar with AWS: "Teaching Claude to be an Engineer" for stateful K8s migrations

https://ku.bz/claude-k8s
December 8, 2025 at 2:06 PM
48 newsletter issues published
The Learn Kubernetes Newsletter now lives on Kube Today—issue 161 drops this week

We hit 80k subscribers

Huge thank you to our sponsors who made this possible!

https://learnkube.com/sponsorships
December 8, 2025 at 2:06 PM
40 podcast episodes of KubeFM!

Season 7 brought amazing guests: GitOps migrations, macOS workloads on K8s, building Kubernetes from scratch in Go…

Thanks Bart for making this happen!
https://kube.fm
December 8, 2025 at 2:06 PM