Choosing Between HPA, VPA, and Cluster Autoscaler in Kubernetes
Guide to choosing and configuring HPA, VPA, and Cluster Autoscaler in Kubernetes based on workload metrics, pod resizing safety, and node‑group elasticity.
Loading…
ReadMeFeed / Community knowledge
Kubernetes
Real questions. Useful conversations. Find the people who know your stack.
Kubernetes
DevOps & observability · #kubernetes
k3OS and the Case for an Immutable Kubernetes Node OS
k3OS stripped the Kubernetes node down to an immutable image and a YAML file. Here's why that design decision matters, how node provisioning actually worked, and what to use now that the project is archived.
k3sDiagnosing Kubernetes CrashLoopBackOff: Exit Codes, Probes & Node Pressure
When a pod enters CrashLoopBackOff, this guide helps interpret exit codes, distinguish probe and scheduling failures, and apply fixes using ordered kubectl checks.
Deploying Dynatrace OneAgent Auto‑Injection in Kubernetes via a DaemonSet
Deploy the Dynatrace OneAgent as a DaemonSet so every node runs the agent, enabling automatic instrumentation of Java processes without code changes. One DaemonSet, a secret for the API token, and DT_TENANT is the minimal design.
dynatraceZero‑Touch Provisioning of k3OS Nodes with a User‑Supplied cloud‑config
Learn how to automate k3OS node bootstrapping using a cloud‑config file so the node joins a k3s cluster without manual intervention.
Choosing a Control Plane Hosting Strategy for Gardener Multi‑Cloud Clusters
Learn how to choose between a managed or self‑managed Gardener control plane to balance operational overhead with compliance needs.
gardenerRolling Updates That Don't Lie to You: Tuning maxSurge, maxUnavailable, and Rollback Depth in Kubernetes Deployments
Kubernetes rolling-update defaults encode assumptions about replica count and cluster headroom that may not match your workload. How to tune maxSurge, maxUnavailable, and revisionHistoryLimit deliberately.
Diagnosing and Fixing TraefikMesh mTLS Certificate Rotation Failures
When TraefikMesh mTLS certs expire after 24‑48 h, services break. This guide walks through symptoms, root causes, step‑by‑step checks, targeted fixes, and when to raise an incident. It covers clock skew, cert‑manager mis‑config, webhook failures, and NetworkPolicy blocks.
traefikmeshWhen Kubeflow Pipelines Earns Its Place (and When a CronJob Wins)
Kubeflow Pipelines pays off when you need caching, lineage, and scheduled retraining on Kubernetes — but a plain CronJob is often the cheaper right answer. Here's how to decide, with a working KFP v2 example.
kubeflow