Page 11 - Linux

When a hundred servers start the same backup job at exactly 00:00, the shared database or the shared storage backend collapses under the load spike, even though each individual job would be harmless on its own. Deliberate scheduling jitter spreads out such start times randomly across a time window and prevents exactly this thundering herd effect. This article shows how jitter is applied in cron and systemd timers.

Docker containers usually run fine in testing, but in continuous operation log rotation, the right storage driver, clean systemd integration and clear resource limits decide whether a single container brings down the whole server or keeps running quietly in the background. This article shows how administrators configure the Docker daemon for production, keep containers starting reliably and continuously monitor their health status.

Solid PCI-DSS hardening requires concrete Linux configuration, not just process documentation. Understanding which requirements of the standard apply directly to the server allows implementing firewall segmentation, access control, logging and file integrity in a way that holds up to a Qualified Security Assessor.

Anyone who cannot start a service because a port is already in use, or needs to know which process is really behind a connection, cannot avoid ss. This article explains why ss replaced the old netstat, how to correctly read connection states like ESTABLISHED and TIME_WAIT, and how to reliably diagnose port conflicts before starting a service.

NTP reliably synchronizes system clocks to within a few milliseconds, but certain applications need far more: sub-microsecond accuracy across the local network. This guide explains how PTP under IEEE 1588 works with hardware timestamping, how linuxptp with ptp4l and phc2sys is set up on Linux, and when the extra effort compared to NTP actually pays off.

Killing production processes with kill -9 risks corrupted PHP-FPM requests, orphaned locks, and inconsistent data. This article explains how Unix signals really work, why SIGTERM and SIGKILL are fundamentally different, how to trap signals cleanly in your own scripts, and how kill, pkill, killall, and systemd work together for controlled process management.

A CIS Benchmark provides the most concrete checklist available for hardening a Linux server traceably and meeting compliance requirements. Anyone who understands the controls, prioritizes them and automates them with CIS-CAT and Ansible replaces manual one off checks with repeatable, auditable processes.

Mount points decide how and where data ends up on a Linux server, and the choice of filesystem determines performance, safety, and recoverability when things go wrong. This article explains the mechanics behind mount, the correct fstab syntax, and shows with concrete examples when ext4, XFS, or Btrfs is genuinely the right choice on a production server.

On multi-socket servers, not every byte of RAM is equally close to every CPU. NUMA describes exactly this asymmetry, and anyone running MySQL or PHP-FPM without regard for NUMA nodes risks silent latency losses that no single application-level setting can explain.

rkhunter and chkrootkit compare known rootkit signatures, file permissions, and process lists against a stored reference state of the system. This guide shows how to install both tools, set them up as a recurring baseline check via cron, and correctly interpret a positive finding without creating a false sense of complete detection.