The advisories don't let up. Every week brings a new batch of CVEs, the kernel alone ships fixes faster than anyone reads them, and the package updates and distribution releases keep coming on top. Keeping a server current used to be a chore you'd get to eventually; now every update you postpone makes the next one a bit riskier.
And there have only ever been two ways out. Most tools set up a server and then leave you with it; managed cloud platforms run it for you, on their infrastructure, under their rules, at their prices. Nothing in between.
strackt is that in between, and as of September it's live. You push a commit and we build, migrate and release it; that's the bit everyone expects, and it works. The part we actually built it for is everything after the deploy: the managed side of the cloud, running on a server you own.
Updates you don't have to schedule
Security advisories are matched against what runs on your servers, approved, tested on one server, then rolled out to the rest. Package and kernel updates arrive the same way, and so do the big ones; a full distribution upgrade, the kind that usually waits for a free weekend, is just another update. Most land without a restart. When one is needed, it happens in the maintenance window you chose, with plenty of notice and a button to push it back; if the bug is being actively exploited, it's sooner.
Configuration that doesn't drift
Every server's configuration comes from strackt, not from whoever last logged in over SSH. Changes are applied in one piece, and a change that doesn't come up cleanly rolls itself back to the last working setup, so a server is never left half-updated. Hand-edits get overwritten on the next update, which is pretty annoying if you like fiddling over SSH; it's also what stops two servers from quietly becoming different.
When it's your code that goes wrong, rolling back a release takes seconds. The server keeps hourly snapshots of your data and applications, and backups go where you point them: another server you manage with strackt, or a bucket you own, encrypted with a key only you hold. You can restore them with standard tools, without strackt in the loop.
Moving things
Moving an application to another server is a flow, not a weekend. We set up the new one and copy the data across while the old one keeps serving; you confirm the switch when you're ready, and if the move fails before then, it rolls back. Databases stay where they are for now, and the application connects to them across servers. WordPress sites can come in from their current host by connecting the live site; that import is still in beta.
What it doesn't do
strackt runs Laravel, plus WordPress, plain PHP, Node.js, Next.js and Nuxt in beta. That's the list; more stacks as they ship.
You bring the server. Any VPS you rent, from whichever provider you like; we manage it, you keep it.
No autoscaling. A server is a server; when you outgrow it, you move to a bigger one.
An environment for every pull request needs GitHub, for now. GitLab, Bitbucket and Gitea get push-to-deploy; previews for them aren't there yet.
It runs code you build. Software you install from someone else's releases (Nextcloud, Immich and the like) isn't what it's for; Coolify is the better tool there.
It's a public beta until the full launch in Q4. No SLA yet; if something breaks, you talk to the person who built it.
The first application is free, and stays free. After that it's €5 per application per month (excl. VAT), flat: no tiers, no usage billing, and your server costs whatever your provider charges.
The advisories will keep coming. That part's our problem now.