June 2020 Summaries

Zac Propersi, an Engineering Manager at Sentry, noticed that the alerts pages were rendering slowly and used a custom query in Discover to see just how slow the transactions were, finding out that an incident's endpoint call was taking seven seconds to complete. The issue was solved by deferring the slow-loading queries to run shortly after the page loads, allowing the most relevant alert data to load first and then trend graphs and supporting statistics would follow, resulting in a landing page that loads in under 300 ms. Application monitoring allows for monitoring all aspects of code, not just errors, enabling Sentry to analyze issues with both frontend and backend feedback loops.

The Sentry project has adopted a new versioning schema called CalVer (Calendar-based Versioning) for its self-hosted releases, replacing traditional SemVer. This change aims to improve alignment between the hosted and open-source versions of the software. With CalVer, Sentry will now ship updates on a monthly cadence, providing more predictable release cycles and reducing the overhead of upgrades. The latest release, v20.6.0, introduces several new features, including Discover v2, Sentry Native, and Release Health, alongside performance improvements and bug fixes.

We use VMware Tanzu Observability (Wavefront) and Sentry to monitor our app infrastructure stack, providing real-time visibility into CPU, Memory, Network I/O, and Storage I/O. We also leverage Wavefront's native integrations with AWS services for resource monitoring and capacity planning. To monitor the Kubernetes stack, we utilize predefined dashboards and integrate Prometheus data using the Prometheus storage adapter. For logging, we use a centralized system to log customer exceptions and errors, which are then processed by Sentry for real-time alerting, exception tracking, and deduplication. Sentry enables us to respond quickly to bugs and exceptions, reducing our time to resolution from weeks to hours, and improving our overall quality of service.

Integrating Sentry with Atlassian products such as Bitbucket, Jira, and Trello aims to help developers spend more time building products and services for their customers by catching software errors before they cause damage. With the new features, teams can get a jump on fixing issues as they ship new features and services, automate processes like creating tickets in Jira or cards in Trello, and escalate critical issues via OpsGenie to notify the right people on their team. This integration enables developers to have all the information they need to reduce resolution time, free up dev cycles, and deliver experiences that create happier customers.

To leverage the full potential of Heroku, integrating it with Sentry provides access to commit data for each release, enabling developers to efficiently debug new errors by tracking changes and identifying team members involved in code deployments. This integration also allows deploy emails to be sent to relevant team members, ensuring timely notification when code is deployed. By connecting these two popular developer tools, users can streamline their development workflow and gain valuable insights into their app's releases.

Netlify has introduced Build Plugins, which allows users to automate and customize their build processes. The Sentry Netlify Build Plugin enables developers to automate release management and deliver a reliable customer experience. To use the plugin, developers need to install it on Netlify, create a new internal integration with Sentry, set environment variables in Netlify, and configure the plugin to notify Sentry of new releases and upload source maps. This automates tedious tasks for developers, unlocking features like suspect commits that likely introduced new errors, applying source maps to see original source code within Sentry, and more.