Daniele Polencic

1/10 One of the things I like about the LearnKube Telegram group is the discussion that follows the articles we share. We shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.

9/10 Even after the downstream service recovers, the containers can still struggle because they compete for CPU. It is also why requests matter beyond placement. They influence how CPU is shared when the node is busy.

1/10 One of the things I like about the LearnKube Telegram group is the discussion that follows the articles we share. We shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.

7/10 The next question was: if the container needs all that CPU at startup, how did it work in the first place? Because it can use spare CPU beyond its request. A request is not a ceiling on CPU usage.

1/10 One of the things I like about the LearnKube Telegram group is the discussion that follows the articles we share. We shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.

5/10 You still only have four CPUs, so the containers compete for them. With equal requests and no other load, each gets roughly 0.4 CPU. The startup work takes longer.

1/10 One of the things I like about the LearnKube Telegram group is the discussion that follows the articles we share. We shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.

3/10 I think this is possible if your requests are tiny, but the actual CPU usage is significant. Or if the container has a huge CPU spike at startup.

1/10 One of the things I like about the LearnKube Telegram group is the discussion that follows the articles we share. We shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.

1/5 This KubeCon NA, we will host our first community event: LearnKube Day in Salt Lake City, on November 9. This one is special to me. For the first time, we will bring our community together for a day of our own. 🧵

4/5 But the part I am most excited about is the chance to meet the people who supported us over the years. If you read our articles, took a course, listened to the podcast, or shared our work, thank you. I hope we can meet in person!

1/5 This KubeCon NA, we will host our first community event: LearnKube Day in Salt Lake City, on November 9. This one is special to me. For the first time, we will bring our community together for a day of our own. 🧵

2/5 To celebrate, Artem Lajko, Diana Todea, Viktor Farcic, Graziano Casto, Brandt Keller, Carol Valencia, and Jay Jackson will join us for lightning talks. Quite a group for our first event!

Agents are becoming part of everyday Kubernetes operations. How do we know when to trust them? Some mistakes produce an immediate error. Others look like successful fixes until production conditions change 🧵

5/6 This is where I get stuck on evals Do platform teams need to maintain incident scenarios alongside their infrastructure to assess their agents? What evidence makes you trust a proposed fix when some mistakes only appear under load or during failover?

Agents are becoming part of everyday Kubernetes operations. How do we know when to trust them? Some mistakes produce an immediate error. Others look like successful fixes until production conditions change 🧵

3/6 OpenAPI specifications, the Flux Schema MCP, versioned documentation, and live telemetry all help If an agent invents a CRD field, validation catches it. The agent reads the error and tries again But what if the answer looks correct?

Agents are becoming part of everyday Kubernetes operations. How do we know when to trust them? Some mistakes produce an immediate error. Others look like successful fixes until production conditions change 🧵

Kubernetes hard mode: how do banks run their platforms? My new book, Kubernetes Architecture in Financial Services, is out now: https://learnkube.com/kubernetes-architecture-financial-services Here is what surprised me while researching it:

Saxo and Itau approached a related problem with operators Saxo reconciled external service dependencies, while Itau used a Kubernetes-style control plane to relocate services across runtimes The overlap is partial, but the contrast is striking

Kubernetes hard mode: how do banks run their platforms? My new book, Kubernetes Architecture in Financial Services, is out now: https://learnkube.com/kubernetes-architecture-financial-services Here is what surprised me while researching it:

Some decisions showed broad agreement. Others led to very different solutions The book covers tenant isolation, delivery, policy, multicluster networking, reliability, service dependencies, and cluster replacement

Kubernetes hard mode: how do banks run their platforms? My new book, Kubernetes Architecture in Financial Services, is out now: https://learnkube.com/kubernetes-architecture-financial-services Here is what surprised me while researching it:

Financial firms have a tougher time with Kubernetes They face the usual problems around scaling, networking, and onboarding, plus regulatory compliance, auditing, and a massive workforce spread across on-premises infrastructure and multiple clouds

When does Kubernetes make sense? It is one of the questions we ask during private Kubernetes courses at @learnkube.com. My controversial answer is one application.

Obviously, if downtime is not an option, this setup does not work for you. But 99.9% availability leaves about 8 hours and 45 minutes of downtime per year. For me, that is plenty for critical updates.

When does Kubernetes make sense? It is one of the questions we ask during private Kubernetes courses at @learnkube.com. My controversial answer is one application.

But if you stick to the core resources and use YAML to describe your infrastructure, Kubernetes is surprisingly manageable. You can do a lot without getting into more complicated scenarios.

When does Kubernetes make sense? It is one of the questions we ask during private Kubernetes courses at @learnkube.com. My controversial answer is one application.

In fact, I prefer very large single-node clusters whenever I can. For example, K3s on machines with more than 128 GB of RAM and 32 cores.

When does Kubernetes make sense? It is one of the questions we ask during private Kubernetes courses at @learnkube.com. My controversial answer is one application.