I've recently integrated Sentry into my full-stack project, using Next.js for the web and Expo for the mobile app, and it's been great for catching errors. However, I'm curious about the workflows different teams adopt for managing Sentry alerts in production. How do you avoid alert fatigue and efficiently handle bugs? I'm particularly interested in: 1. **Workflow & Alerts:** How do you filter out unnecessary alerts? Do you keep development and production environments separate, or do you set up smart alerts that notify you only when a bug surpasses a certain threshold? 2. **Automated Bug Fixing:** With the rise of AI coding tools, I've heard of a developer using a custom script with Claude to fetch open Sentry errors and generate code fixes using an LLM. Is anyone employing similar methods? Are you creating your own LLM scripts or using existing tools like Sentry's AI features or Sweep.dev? I'd love to learn how others transition from identifying bugs to fixing them, especially with automation involved!
1 Answer
We keep our Sentry sample rate at 0.02, focusing only on the standout issues during a monthly review or recurring errors. We're still doing manual fixes because it helps us understand the problem correctly, especially with important things like layout shifts or search engine data structures.
That low sample rate is interesting! I’m planning to launch a new codebase soon, so I might start higher to catch critical issues early. Monthly reviews for less critical errors sound smart. Thanks for the reminder about the importance of manual fixes for layout problems!