Page 11 - Linux

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.

A backup script that works fine most of the time but leaves an incomplete archive behind when disk space runs out, one that nobody recognizes as broken, is worse than no backup at all, because it creates false confidence. Robust tar and gzip automation means handling failure cases just as carefully as the success case.

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.

OpenVPN and IPsec have been the standard for site to site VPNs for years, but they carry a code footprint, configuration complexity, and handshake overhead that WireGuard deliberately avoids. With roughly four thousand lines of kernel code, modern cryptography, and a single configuration file, WireGuard has established itself as a lean alternative for secure server to server connections.

Once more than a single IP address needs managing, for example a filesystem, a database and a service all becoming resilient together, keepalived is no longer enough. Pacemaker and Corosync together form the standard Linux cluster stack that monitors, starts and relocates arbitrary resources in a coordinated way on failure.

When development environments, databases and Docker containers live on a remote server, the quality of the SSH connection decides whether the work feels local or like a constant compromise. With a clean SSH config, VS Code Remote-SSH and targeted port forwarding, the remote server becomes a seamless extension of your own development environment.