Cache warmup bug caused hourly database crashes after every deploy

·1 views

A product catalogue service began experiencing six-minute outages roughly an hour after each deployment, with database CPU maxing out and page loads spiking to eight or nine seconds. The root cause was a well-intentioned cache warmup step that preloaded 180,000 product entries with identical one-hour TTLs before each pod went live. Because all entries were loaded at the same time, they all expired simultaneously, triggering a flood of database queries the system was not sized to handle. The original lazy-loading cache had naturally spread expiry times through organic traffic, providing accidental protection that engineers unknowingly removed. The team resolved the issue by adding random TTL jitter, coalescing duplicate cache-miss requests, serving stale content during background refreshes, and isolating cache-fill database connections.

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)

OGAD Tool Lets Windows Users Summarise Long Documents Without Cloud AI

OGAD (Off Grid AI Desktop) is a Windows application that allows users to summarise lengthy documents using a locally run AI model, eliminating the need to upload files to cloud services. The tool supports text-based formats including PDF, DOCX, TXT, and Markdown, though scanned PDFs require extractable text before they can be processed. Users are advised to work through documents section by section rather than posing a single broad question, checking each AI-generated answer against the source before combining notes into a final summary. The app's project and document search features are available as free core functions, requiring a compatible Windows x64 machine with sufficient memory for the chosen local model. Because very large documents may exceed the import size limit, lengthy files may need to be split before indexing to ensure complete coverage.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Simple invisible-character stripping can break emoji while letting hidden LLM prompts through

A technical analysis warns that naively removing invisible Unicode characters from text can destroy legitimate emoji — such as family sequences built with Zero Width Joiner characters — while failing to strip hidden instructions embedded via Unicode tag characters. The article identifies four distinct families of invisible code points, each with different legitimate uses and different security risks, arguing that treating them as a single category leads to both over-removal and under-removal. Unicode defines a formal set called Default_Ignorable_Code_Points, covering 4,174 characters, which the author uses as the correct detection baseline rather than short hand-written regex patterns that typically cover only a handful of these. The recommended approach assigns a separate policy to each family — keeping joiners and bidi controls in prose, but removing tag characters entirely, as they have no legitimate rendering use and can carry hidden text instructions to LLM applications. The piece also notes that visually blank but width-bearing spaces like U+202F, reportedly appearing in OpenAI model output, fall outside this set and require a separate whitespace normalisation policy.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

OGAD App Lets Mac Users Summarise Long Documents Using Local AI, No Cloud Needed

A Mac application called OGAD (Off Grid AI Desktop) allows users to summarise lengthy documents using a locally run AI model, without uploading files to any cloud service. The tool supports text-based PDF, DOCX, TXT, and Markdown files, and requires an Apple Silicon Mac running macOS 13 or later. Users create a project, import a document, and query individual sections through a chat interface to build a verifiable, section-by-section summary. Because the app does not process every page per query, very large documents may need to be split before indexing, and scanned PDFs require extractable text to work properly. The recommended workflow involves checking each section's output against the source before combining reviewed notes into a final summary.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

File Size Does Not Determine Image Quality — Here Is What Actually Does

A common misconception in digital imaging is that a larger file size guarantees better visual quality, but this is not the case. File size simply measures how much data is stored, not how much useful visual detail an image actually contains. Two images with identical pixel dimensions can differ vastly in file size depending on their compression settings, format, and content complexity. A simple graphic with smooth color transitions can look sharp at a small file size, while a poorly compressed large file may appear blurry or blocky. Ultimately, image quality depends on how efficiently a file encodes its visual information, not on the raw number of kilobytes or megabytes it occupies.

0 comments Read more at DEV Community