1732 posts tagged with "Self-Hosting"
Running your own PaaS and infrastructure on machines you own
Coolify v5 Says No to Kubernetes: What 'Like Swarm but Better' Means at Box Two
Coolify v5 bets on a custom Compose-native orchestrator instead of Kubernetes. A six-capability box-two scorecard — rolling deploys, scheduling, storage, networking, machine lifecycle, conformance — with evidence tags on every v5 promise, plus the four signals that tell you whether to wait or leave the single-box ceiling now.
Coolify vs Dokploy After a Real Migration: What the Same Laravel App on Two Identical Hetzner Boxes Reveals
The same Laravel app plus Postgres, deployed to Coolify and Dokploy on two identical Hetzner CPX21 boxes: install times, Railpack-vs-Nixpacks build speeds, and the one-click-vs-CLI restore gap that picked the winner — plus where the result confirms our second-machine showdown.
No Patched Kernel Exists Yet: What CVE-2026-80521 Costs a Self-Hosted Fleet During the Patch Gap
CVE-2026-80521 escapes default containers to host root with a public exploit and no fixed Ubuntu or Debian kernel — a four-step runbook to check exposure, triage untrusted workloads, sandbox through the gap, and roll the fix.
Daytona Raised $24M in February, Then Closed Its Source in June: A Buy-vs-Build-vs-Fork Read on Agent Sandboxes
Daytona's $24M Series A and June closed-source pivot, E2B's Apache-2.0 runtime, Modal's $355M scale bet, and Runloop's seed — a vendor scorecard plus the DRA, snapshot, and concurrency work you inherit when agent sandboxes move onto owned hardware.
Leaving Heroku With Django? Inventory First, Price Table Second
A Django-on-Heroku migration inventory — buildpack pins, release-phase migrations, collectstatic, Celery topology, and Postgres extensions — that decides whether your exit to Render, Railway, Fly.io, or owned hardware is a weekend or a quarter, before any price table can tell you.
Flipt v2 Goes Git-Native: Feature Flags That Branch With Your Code
Flipt v2 stores feature flags in Git instead of a database, turning releases into reviewed pull requests. A comparison of the 2026 self-hosted flag field plus a worked deploy-dark workflow for a git-push PaaS, with the license and Pro-gate caveats spelled out.
Fly.io's Edge Premium, Priced: A Single-Region Always-On App vs Render, Railway, and a $7 Hetzner Box
Fly.io's $2.19 headline assembles to $19 for an always-on single-region stack — priced line by line against Render (~$34), Railway (~$57), and a $7 Hetzner box, with an egress sensitivity table and the honest carve-out where the edge earns its keep.
Fly.io's Logs Went Dark for 4.5 Hours While Apps Stayed Up: What Owning Your Log Pipeline Actually Costs
Fly.io's historical logs lagged 4 hours 24 minutes on October 13 — its second log-ingestion incident in thirteen days — while apps kept serving. This teardown maps which half of the pipeline failed, prices 10 GB/month of logs across Datadog, Better Stack, Axiom, and a €4.35 self-hosted VictoriaLogs box, and shows how owning the log path turns the next backed-up queue into your own alertable metric.
Fly.io's 37-Region Edge vs the Two Regions You Actually Serve: Pricing the $3.88 Machine Against the $7 Heroku Dyno
Fly.io's $3.88 shared-cpu machine undercuts Heroku's $7 dyno per process — but the assembled two-region stack costs $30 vs $23 once Postgres enters, against ~$5–10 on flat Hetzner hardware, plus the edge ops overhead neither unit price shows.