Linux

No space left on device in the middle of live production is one of the most unpleasant alerts there is, because MySQL, PHP sessions and logging can all fail at once. This article delivers the complete emergency workflow for a full disk: safely and quickly reclaim space, identify the actual largest consumer, and then make sure the incident does not repeat itself.

Anyone who wants to open a single graphical tool on a remote Linux server does not need a full remote desktop. X11 forwarding over SSH is usually entirely sufficient for that. For a complete desktop with several windows and a persistent session, a VNC server is instead the far more robust solution. Both approaches have clear use cases that can be separated cleanly based on latency, security and application scenario.

A failover mechanism that has never been tested under realistic conditions is only an assumption about resilience, not proof of it. Systematic failover testing with controlled failure simulations, clear metrics like RTO and RPO, and a repeatable test plan uncovers exactly the gaps that would cause long outages in a real emergency.

Where silent data corruption is not an option, a classic filesystem paired with a RAID controller often falls short. ZFS combines volume management and filesystem into a single layer, checksums every block, and repairs corrupted copies automatically whenever redundancy exists. This article covers installation, pool setup, and a practical example for database storage.

Name resolution failure is one of the most common and, at the same time, one of the least understood errors in Docker setups. Anyone who knows how the embedded DNS server works and which tools expose the root cause can find DNS issues in Docker networks in minutes instead of hours.

Servers that run SSH with password login and an open root account are a preferred target for automated brute force attacks from across the internet. This article shows in practice how key based authentication, a consistently hardened sshd_config, clean client configuration through host specific entries, and fail2ban as a complementary layer secure SSH access on production Linux servers reliably and permanently, without exotic extra tools.

Between loading the kernel and mounting the actual root filesystem sits a step that stays invisible on most servers until it fails: the initramfs. Anyone running LVM, LUKS, or a custom storage setup needs to understand how dracut builds this minimal pre boot system, so that an unable to mount root error does not turn into hours of troubleshooting.

SUID, SGID, and the sticky bit control which rights a program runs with, who inherits group ownership, and who is allowed to delete files in shared directories. Used correctly, they enable safe system tools like passwd and stable shared project folders without constant manual permission fixes. Misconfigured, unnoticed SUID binaries open a classic path to privilege escalation that every administrator should audit regularly.

System monitoring starts with correctly interpreting load average, CPU percentages, and memory figures in top and htop, because misreadings lead to wrong diagnoses. This article shows how to sort and filter process lists to find the actual cause of a problem, when modern alternatives like btop and glances are worth the extra install, and what a structured workflow looks like during an acute live incident.

Server logs almost always contain personal data such as IP addresses, which makes any GDPR log retention a concrete legal question. Anyone who knows the retention periods, configures logrotate correctly and implements pseudonymization technically operates logging that remains both forensically useful and privacy compliant.