Journal
Engineering notes, no fluff.
Architecture, security, incident response, AI kept on a short leash. We write the things we wish we had read before building Servor.
Archive
02 — 15- October 5, 202610 min read
Running Redis on your own server: cache, persistence, and the port you must not leave open
Redis is one command to install and two completely different databases to run. Choosing which one you want — and closing the port before an attacker does — is the whole job.
Read - September 28, 202610 min read
Caddy and automatic HTTPS: the certificate you never have to think about again
Caddy gets you a valid certificate in one line and renews it forever. The one line is easy. Knowing what has to be true before it works is the part worth reading.
Read - September 24, 20269 min read
XPipe vs Servor: connection hub or server console?
XPipe is the best connection manager we have used. It is also a client on one laptop, which is exactly the boundary Servor is built to cross: monitoring while you sleep, execution that survives a vendor breach, a fleet your teammates can operate too.
Read - September 21, 202611 min read
A Prometheus and Grafana stack you can stand up in an afternoon
Everyone wants dashboards. Fewer understand that Prometheus scrapes, node_exporter exposes, and Grafana only draws — and that the default that ships is wide open. Here is the stack, honestly.
Read - September 14, 202611 min read
Installing Docker on a server without leaving a root-shaped hole
Installing Docker is one command. Understanding that you just handed out root, and deciding to whom, is the part that separates a server from an incident.
Read - September 10, 20269 min read
Ploi vs Servor: deploying sites or operating a fleet
Ploi answers "how do I get this site live and redeploy it without downtime". Servor answers "what are my servers doing right now, who gets told when one breaks, and how do I fix it from wherever I am". Most agencies have both questions.
Read - September 7, 202611 min read
PostgreSQL in production: install, harden and back up without losing the cluster
Installing Postgres is ten minutes. Running one you can trust — the right version, an access policy that is not wide open, and a backup you have actually restored — is the part that earns the word production.
Read - September 2, 202612 min read
The best server management tools in 2026: Servor, XPipe, Cockpit, Webmin, Termius and more
“Server management tool” covers six different jobs. We rank the best option for each — XPipe, Cockpit, Webmin, Termius, Uptime Kuma, Portainer, Ansible and Servor — in one feature matrix with ticks, crosses and the red crosses Servor earns too.
Read - August 31, 202610 min read
Putting an app behind nginx: the reverse proxy, done without breaking the machine
A reverse proxy is three files and one reload — and every one of them has a way to take the site down. Here is the shape of the job, and the checks that keep an existing configuration intact.
Read - August 19, 202610 min read
Reducing alert noise: being slow to shout is a feature
A pager that fires every night stops being read, and then the one real outage arrives on a channel nobody looks at. Here is how to tune detection so that what survives is what a human can act on.
Read - August 12, 202611 min read
Hardening SSH on a fresh server: most guides optimise for the wrong threat
A hardening checklist is easy to follow and easy to get wrong. Here is what actually moves your exposure on a machine you were handed an hour ago — and what is decoration.
Read - May 12, 20268 min read
Why we built Servor around Plan, Execute, Verify
Nobody can afford to let an AI agent run blind on production servers any more. Here is the three-beat discipline that shaped Servor's entire architecture.
Read - May 5, 202611 min read
Zero-knowledge in practice: how Servor protects credentials it can never read
Everyone says "zero-knowledge". Very few products actually are. Here is how Servor is wired so that even a full database breach leaves your credentials unreadable.
Read - April 22, 20269 min read
Status pages that don't lie: a 2026 guide to incident communication
A green status page is not an honest status page. Here is the practical guide to turning yours into an instrument of transparency rather than a fig leaf.
Read