Three Commands to Diagnose and Free Up Disk Space on Linux

·1 views

A full disk is one of the most common causes of unexpected service failures on Linux systems. Administrators can identify which partition is full using the 'df -PhT' command, which provides a clear overview of disk usage across all partitions. The largest space-consuming directories can then be pinpointed by running a 'du' command that lists the top 10 folders by size. Quick cleanup options include truncating large log files, clearing the package cache with 'apt clean', and checking for deleted files still held open by processes using 'lsof'. These three steps form a practical checklist for resolving low-disk-space issues without advanced tools.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Free animated platform teaches data structures with step-by-step visuals, no signup needed

A developer has launched learningto/pass, a free interactive platform designed to teach classic data structures and algorithms through step-by-step animations rather than static diagrams or rote memorization. The tool covers 16 topics ranging from Big O notation to dynamic programming, with each concept visualized before any code is introduced. Python runs directly in the browser via WebAssembly, requiring no installation and working on most devices including phones and Chromebooks. Practice sections offer interview-style coding challenges with AI feedback that mimics a mock interviewer, providing complexity analysis and progressive hints rather than direct answers. No account is required to start, and progress syncing across devices becomes available upon sign-up.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer built localization guardrails first, yet three defects still slipped through

A developer adding Russian and Spanish support to Parlotype, a local voice-to-text desktop app built on .NET 10 and Avalonia, used AI coding agent Claude Code to extract and translate roughly 450 string keys across 122 files. Before any bulk extraction began, the developer deliberately built the quality guardrails first — including a parity checker script, xUnit tests, Claude Code hooks, and a skill file — to catch errors automatically during the agent's work. The approach was chosen specifically because AI agents doing mechanical bulk tasks tend to drift, and catching mistakes inside the loop is cheaper than catching them at review time. Despite all these safeguards, three user-visible defects still shipped to the final build, none of which were translation errors. The post-mortem examines each defect as a distinct category of blind spot that the guardrail system failed to anticipate.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How GeoBlood Uses MongoDB Geospatial Queries to Match Blood Donors in Emergencies

GeoBlood is a blood donor registry platform that uses MongoDB's geospatial indexing to match nearby compatible donors during medical emergencies. A naive approach of fetching all donors and filtering in application memory fails at scale, scanning entire collections and wasting CPU and memory resources. The system also accounts for blood type compatibility rules, such as O-negative donors being eligible for B-negative recipients, which a simple query would miss. Additional factors like recent donation history and donor responsiveness must be modeled to avoid routing requests to ineligible or unreliable donors. The project also highlights a critical GeoJSON pitfall where longitude and latitude must be stored in reverse of the common human convention, with incorrect ordering silently returning wrong results rather than throwing errors.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Are Developers Getting Addicted to AI Coding Tools Like Gamblers?

A software developer has drawn parallels between the growing use of AI coding assistants and gambling addiction, arguing that the cycle of prompting and receiving results mimics the behavioral patterns of placing bets. The author observes that instead of saving time as promised, AI tools appear to be driving developers to work longer hours and spend more money on escalating subscription plans. Psychological mechanisms such as the near-miss effect and variable reward responses are cited as reasons why users keep pushing harder rather than stepping away. The author raises concerns that this compulsive pattern may lead developers to take greater risks, reduce scrutiny, and ship lower-quality code faster. The piece was originally written in June and republished on DEV Community as the author began engaging with the platform.

0 comments Read more at DEV Community