1732 posts tagged with "Self-Hosting"
Running your own PaaS and infrastructure on machines you own
What KubeCon Amsterdam Disagreed About: Golden Paths, Hidden Kubernetes, and Agents as Users
KubeCon EU 2026 Amsterdam agreed platform engineering is the standard org model but split on golden-path scope, Kubernetes abstraction depth, and agents as platform users. This synthesis maps the three fault lines with verdicts for small self-hosted teams.
KubeCon NA 2026: What a Solo Self-Hosted Operator Should Actually Watch in Salt Lake City
KubeCon + CloudNativeCon North America 2026 runs November 9-12 in Salt Lake City with a new AI Inference + Agentic track and eleven Monday co-located events. This filter gives a two-person self-hosted fleet team six prioritized picks — named sessions, the Monday pass decision, maintainer-track tactics — plus a skip list and a free remote-follow playbook.
The Most Useful Date in Kubernetes 1.37 Was July 22, Not August 26
Kubernetes 1.37's freeze calendar — enhancements freeze June 17, code freeze July 22, GA August 26 — gave fleets ten scope-locked weeks to audit cgroup v1 nodes, containerd 1.x, and deprecated kubelet flags before upgrading. Here is the milestone-to-action runbook mapping, worked against the real 1.37 breaking changes.
Kubernetes' AI Gateway Working Group: Build the Gateway Boundary Before You Add Another MCP Tool
Kubernetes' AI Gateway Working Group treats agent traffic as an operational boundary, not another app route. Here is the four-control gateway — agent identity, quotas, tool-call tracing, deny-by-default egress — a self-hosted PaaS should build before exposing deploy and rollback tools over MCP.
Your API Server Already Speaks Post-Quantum TLS — Your Service Mesh Probably Doesn't
A TLS-hop inventory and 4-phase migration playbook for a Cluster-API fleet: which hops already negotiate hybrid ML-KEM, what Envoy, Istio, and Linkerd each need, and where ML-DSA certificates stand.
Gang Scheduling Is Beta in Kubernetes 1.37 — and Off: The Opt-In Checklist for GPU Node Pools
Kubernetes 1.37 graduates gang scheduling to beta — disabled by default — so a bare Garhwal upgrade changes nothing about GPU placement. Enable GenericWorkload on all three control-plane components, drop the removed alpha gate, wire PodGroups or Job scheduling policies, and roll it out per cluster: the full opt-in checklist for CAPI-managed GPU pools.
KYAML Goes Stable in Kubernetes 1.37: What Canonical kubectl Output Means for GitOps Diff Hygiene
Kubernetes 1.37 graduates KYAML to stable, giving kubectl a canonical YAML output that collapses reformatting noise out of GitOps diffs — normalize both sides before comparing, and keep Flux drift detection for the semantic layer it can't see.
Let's Encrypt's 64-Day Certificates Are Live in Staging — Your Renewal Automation Has 4 Months to Prove Itself
Let's Encrypt staging now issues 64-day certificates ahead of the February 10, 2027 production cutover — two days after a 6-hour ACME outage. This sequel to our cert-manager audit delivers the 4-month proof plan: kill hardcoded renewal offsets, switch to ARI scheduling, run staging-canary renewals, and line up a fallback CA.
Your AI Agent Has Shell Access. Lima 2.1 Builds It a Blast Room.
Lima 2.1's sterile --mount-none boot plus the --sync review gate contain an agent's worst command inside a VM copy you approve before it touches your tree. This teardown walks the exact workflow, the v2.0 AI-inside vs MCP patterns behind it, its honest limits, and why hardened local VMs plus per-branch previews on owned hardware beat hosted dev environments.