Dawson’s first law of computing strikes twice.

Sentry is a platform for monitoring, troubleshooting, and error tracking in software applications. Readers can learn about logging, error handling, and performance monitoring techniques. With documentation, tutorials, and best practices, Sentry helps developers identify and fix issues in their applications, ensuring a seamless user experience.

Sentry Engineering

Two production incidents at Sentry were caused by O(n^2) algorithms that passed testing but failed with larger datasets. The Android case involved DEX file symbolication where a linear scan fallback created quadratic complexity with 52,202 classes, solved by adding a dictionary lookup. The iOS case involved LIEF library's export trie parsing with an unpopulated cache causing linear scans, fixed with a simple patch. Both demonstrate how O(n^2) algorithms are fast enough to reach production but slow enough to cause failures at scale.

Next Halloween I’m dressing as O(n^2)