The engineering blog
How we build and run web applications: databases, deployment, scheduling and the small decisions that keep a site simple.
The PHP2Shell update closes the hole, but attackers were already trying it before the fix came out. What to ask your developer or host, which checks cover the days before the patch, and what to do if something looks wrong.
Bozhidar Ivanov
When the person who built your software is gone, starting again can feel like the safe choice. Secure the accounts first, test a restore and a deploy, and let what you find decide between rescue, repair and rebuild.
Bozhidar Ivanov
A Hetzner server accepted its remote disk unlock but stayed offline, because early boot network state named an interface the running system no longer used. See the evidence, the repair and the reboot checks.
Ivaylo Valchev
We converted one existing Hetzner Cloud root disk to LUKS2 without reinstalling. Settle the boot path, partition space, early unlock, and recovery checks before you start yours.
Ivaylo Valchev
Five deployment problems that pass every local test but only surface on a real server, and the habit that catches each one before it reaches users.
ArcAI Team
Scheduled posts usually mean a new status and a cron job. A future date field can do the whole job, if you write down what's allowed to catch up late.
Bozhidar Ivanov
SQLite can run a real production site with several writers, provided you set three things up correctly. The quick version, and the signs that mean it's time to move on.
Bozhidar Ivanov
