Coolify v4 and the v5 Rewrite: What 325,000 Users on a Single-Box PaaS Say About Multi-Server Scalability
Coolify v4 polished the single-box self-hosted PaaS to a shine — then announced v5 as a ground-up rewrite for multi-server scalability. A mechanism-by-mechanism look at why fleet behavior can't be bolted onto per-host SSH orchestration, and what a Kubernetes/Cluster API platform inherits on day one.
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.
Kubernetes 1.36 Lets You Resize a Job Before It Starts: The Queue-Then-Resize Pattern a Build Queue Needed
Kubernetes 1.36 lets a queue controller patch a suspended Job's CPU/memory/GPU requests down to whatever's actually free, then unsuspend it, without losing the Job's identity or history. Here's the worked example, the safety guarantee behind it, and how to wire it into a build-queue controller.
The Self-Hosted PaaS Category Gets Crowded in 2026: What Coolify, Dokku, CapRover, and Dokploy Actually Don't Do
Coolify, Dokku, CapRover, and Dokploy now cover the self-hosted PaaS category from every angle. A grounded look at the three things none of them do — and the real cost of the Kubernetes-native alternative that does.
Coolify, Dokku, and CapRover Solve Git-Push Deploys on One Box — Here's Exactly Where That Wall Is
Coolify, Dokku, and CapRover all run one Docker daemon as the control plane, build runner, and runtime. Here are the actual RAM, CPU-during-deploy, and reboot-failure numbers that mark where that single box stops working, and what a Cluster API fleet does differently.
vCluster Brought Karpenter to Bare Metal. Here's What Actually Changes for a Cluster-API Fleet
vCluster's Auto Nodes brings Karpenter-style provisioning to bare metal. Here's the concrete mechanics, the Hetzner Cloud-vs-dedicated reality check, and the per-tenant control-plane cost a Cluster-API fleet actually pays for it.
Kubernetes Swap Went Stable in 1.34: What LimitedSwap Actually Buys a Self-Hosted PaaS on NVMe
Kubernetes' node swap support graduated to stable in 1.34. A walkthrough of what LimitedSwap actually protects against, what it doesn't, and the concrete tenant-density math it unlocks on a Hetzner NVMe node that a fixed-instance managed PaaS can't touch.
SNCF Cut Kubernetes Provisioning From a Month to 30 Minutes With Cluster API
France's national railway runs its own datacenters on Cluster API, cutting cluster provisioning from a month to 30 minutes and incidents by 90% — here's what that proves about the CAPH/CAPD provider model bex is built on.
Kubernetes 1.35 Takes In-Place Pod Resize to GA: What the Restart-Based Workaround It Just Killed Was Actually Costing You
In-place pod resize reached GA in Kubernetes 1.35. Here's the before/after: what evicting and rescheduling a pod to change its CPU/memory actually cost versus a resize subresource PATCH that completes in seconds — and what a Cluster-API-managed PaaS's own autoscaling logic should do differently now.