1732 posts tagged with "Self-Hosting"
Running your own PaaS and infrastructure on machines you own
Kubernetes 1.36's Memory QoS Goes Tiered: What TieredReservation Actually Guarantees on a Multi-Tenant Node
Kubernetes 1.36 lets a kubelet write hard memory.min protection for Guaranteed pods and soft memory.low for Burstable ones. Here's exactly what that stops a noisy neighbor from doing on a bin-packed multi-tenant node — and the two failure modes it still doesn't cover.
Kubernetes 1.36 Didn't Put DRA in GA — Here's What Shipped, and How to Run GPU-Aware Scheduling on Your Own Hetzner Fleet
Kubernetes 1.36 didn't put Dynamic Resource Allocation into GA — that happened in 1.34. Here's what actually shipped for GPU scheduling, and how to get a real DRA-scheduled GPU node onto a self-hosted Cluster API fleet on Hetzner.
Kubernetes 1.36's HPA Can Now Scale to Zero: What It Actually Changes for Idle-App Billing
Kubernetes 1.36 finally defaults HPAScaleToZero on, but it only solves half the scale-to-zero problem. Here's the exact split, real cold-start numbers from Render and Fly.io, and a worked cost model for a self-hosted PaaS's dormant-app billing.
Kubernetes 1.36's PSI Metrics Graduate to GA: What Pressure Stall Information Catches About Resource Starvation That CPU% Utilization Misses
Kubernetes 1.36 locks PSI metrics to GA, catching resource stalls a CPU% dashboard can't see. Here's the worked example, the two ways the signal can lie, and the pre-flight checklist a bare-metal fleet needs before trusting it.
KubeVela's OAM Delivery Engine Runs Alibaba's Hybrid Cloud and ByteDance's Gaming PaaS: What It Buys Over a Simpler Git-Push Deploy API
KubeVela's component-trait-policy-workflow model runs Alibaba's hybrid-cloud fleet and ByteDance's gaming PaaS. Here's the concrete YAML gap between that and a Render-style git-push deploy API — and when the extra layer is worth the complexity.
Your Deploy Agent's Sandbox Will Die Mid-Rollout — Here's How Its MCP Server Should Survive That
MCP's July 2026 spec update made the protocol stateless — which means the burden of surviving a crashed sandbox now falls entirely on the MCP server. Here's the Postgres checkpoint design, borrowed from LangGraph, that a deploy-from-chat PaaS needs to make a multi-step rollout outlive the agent that started it.
Designing upgrade_cluster: An MCP Tool Contract for Fleet-Wide Kubernetes Node Upgrades
A deploy/rollback MCP tool that goes wrong costs a bad release. An upgrade_cluster tool that goes wrong can leave a fleet straddling a Kubernetes version-skew boundary mid-drain. Here's a concrete tool contract — plan mode, per-node health gates, and a hard stop — for handing an agent that authority safely.
The Rollback Floor: Designing a Deploy-From-Chat MCP Server That Can't Roll Back Too Far
A dashboard's rollback button only ever offers 'go back one.' An MCP tool wrapping the same API can't assume that restraint — here's a concrete tool contract, server-side floor, and audit schema that stop an agent from rolling back past the last known-good release.
MCP's '10,000 Connections at Sub-50ms' Claim Is True — and Doesn't Mean What You Think
The 2026 benchmark headline about MCP servers handling 10,000 connections at sub-50ms is technically true and mostly beside the point. Here's what a production deploy-from-chat MCP server actually needs when a fleet of agents, not one chat session, is calling it.