Page 2 - Monthly Archives: May 2025

MySQL 8 introduced two small but practically important index features that were simply missing in earlier versions: true descending indexes and invisible indexes. Before, the optimizer simulated a descending sort with a backward scan over an ascending index, which hit a hard wall once mixed sort directions across several columns were involved. And an existing index could only be tested by actually dropping it, a risky step in a production Magento database. This article covers how both features work internally, when rebuilding an existing index is genuinely worth it, and how to make index changes to a live installation with minimal risk.

Redis has no native concept of multi-tenancy out of the box, which is why every multi-tenant setup requires a deliberate architectural decision. Three approaches are typically on the table: prefix-based namespaces within a shared database, logical databases via the classic SELECT index, or fully separate instances per tenant. All three differ significantly in isolation level, backup granularity, and scalability. This article compares the three strategies against concrete criteria and derives a practical recommendation for multi-store Magento setups, where several store views are meant to share the same cache layer.

A video embed without a fixed aspect ratio shifts the entire layout while it loads, and it costs Core Web Vitals points. Tailwind CSS's aspect-ratio utilities solve this problem cleanly, responsively and without a padding hack. This article shows every pattern, from 16:9 to fully custom ratios.

Log files that are never rotated grow silently until a disk fills up and a production server crashes. logrotate keeps log files in check by size or time, compresses old files, and runs hooks such as service restarts after rotation. This article shows a practical configuration for a custom PHP application log and how to test it safely with debug mode before going live.

Green tests do not mean the code is safe. In many teams, tests only get a superficial look in review, whether they run, whether coverage numbers add up. Yet it is entirely different criteria that decide whether a PHPUnit test provides real protection against regressions or merely fakes safety.

A modular monolith splits a Symfony application into clearly bounded modules, each shipped as its own bundle with its own data layer and defined interface. That keeps a growing project maintainable without immediately taking on the operational complexity of microservices. This article shows the concrete directory layout, the communication between modules and how Deptrac enforces architectural boundaries automatically.

The INFO command delivers a wealth of raw data about a Redis instance, but as plain text output it is barely usable for systematic alerting. The Redis Exporter translates exactly this raw data into structured Prometheus metrics, extends it with custom check commands for key sizes and latencies, and only then makes it usable for dashboards and alerting rules. This article covers which metrics the exporter concretely provides, what sensible alerting rules for memory usage, connection count, and replication lag look like, and how to build a dashboard for daily operations from it that fires not only on an acute outage, but already at the first warning signs.

Every call to clock_gettime, every timestamp generated by MySQL, and every latency measurement in PHP-FPM depends on the Linux kernel's active clocksource. The Time Stamp Counter (TSC) is orders of magnitude faster than HPET on modern hardware, but was long known for synchronization problems across CPU cores that still produce false expectations and misconfigurations today.

Checkboxes, radio buttons, range sliders and progress bars carry a different default color in every browser, usually some system blue that either matches a brand or does not. The accent-color CSS property solves this with a single line, without native keyboard control, focus rings or screen reader semantics ever getting lost.

Mastodon and the fediverse behind it work fundamentally differently from centrally controlled social platforms: thousands of independent instances communicate with each other through the open ActivityPub protocol, each with its own robots.txt configuration and its own rules. For SEO practitioners this means inconsistent indexability, but also real opportunities, especially for companies with a tech savvy B2B audience. This article explains the technical mechanics, the pitfalls of brand presence spread across multiple instances, and the practical relevance for developer and open source adjacent audiences.