Page 6 - Linux

A server that sits in Berlin but is configured with an American time zone produces timestamps nobody can interpret without conversion. timedatectl sets and checks time zones centrally under systemd, and tzdata supplies the underlying rule database behind it. This article shows how to set time zones correctly, why servers should mostly run on UTC, and how tzdata updates work.

fail2ban watches log files from SSH and web applications in real time, detects repeated failed logins using configurable patterns, and automatically bans the offending IP address at the firewall. This article shows how administrators can correctly configure jails for SSH and web logins, tune ban times sensibly, and monitor fail2ban activity on an ongoing basis without accidentally locking out legitimate users.

APT manages software packages on Debian and Ubuntu servers, yet most admins use only a fraction of its capabilities. This article explains how APT resolves dependencies, why apt, apt-get, and dpkg serve different purposes, how pinning prevents unwanted upgrades, how to safely add third-party repositories, and how to keep the package cache clean.

Too little swap leads to crashes from the out-of-memory killer during load spikes, too much swap hides a real memory shortage behind agonizing slowness. This article shows how much swap actually makes sense on a modern server, how the vm.swappiness kernel parameter controls swapping behavior, how to safely create a swapfile, and how to tell real memory pressure apart from normal caching.

A classic cron job scheduled to run at 03:00 gets simply skipped on a powered off laptop, without any notification. Anacron solves exactly this problem by checking, at the next power on, which jobs were due and automatically running them. This article shows how anacron works, how to configure /etc/anacrontab, and how to sensibly combine it with classic cron.

"Segmentation fault (core dumped)" in the log marks the end of the investigation for many teams instead of the beginning. Yet a correctly enabled core dump combined with gdb and coredumpctl almost always yields a complete backtrace down to the exact line of code. This guide shows how to configure core dumps correctly on Linux, read backtraces, and systematically narrow down typical causes such as faulty PHP extensions.

A single application server does not scale indefinitely and goes offline instantly on failure. HAProxy distributes requests across multiple backend servers, automatically detects failed instances through health checks, and turns several individual servers into a resilient, horizontally scalable unit.

Not every scheduled task has to be recurring. For one-off actions at a specific point in time, or dependent on system load, at and batch are the leaner tools compared to cron and systemd timers. This guide explains syntax, time expressions, job management with atq and atrm, and permission control through at.allow and at.deny.

Installing the same tool via apt one day, pipx the next, and a version manager after that eventually produces conflicting versions and hard-to-reproduce bugs. This article explains when system-wide installation, user-local installation and containerized tools each make sense, and provides a clear policy for shared team servers.

Anyone who only knows systemd through systemctl start and systemctl status is seeing just a fraction of the init system. Target units structure the boot process as a directed graph, socket units start services only on the first connection, and both concepts work together with timer units to save resources and model dependencies cleanly.