August 2019 Summaries

Uptime is defined as the time when an application or service is operational, with operational meaning varying across different contexts. Measuring uptime typically uses nines, where each nine represents a level of reliability, with one nine equating to 36.5 days of downtime per year and five nines representing less than 6 minutes of downtime per year. Achieving higher nines requires more validation and automation, as well as a carefully cultivated engineering culture that encourages high quality and accountability. Balancing error prevention and awareness is crucial for improving service uptime, with code validation and checks contributing to cleaner code that prevents errors in production. Application monitoring provides real-time insights into a service's health, alerting developer teams to issues immediately and minimizing the mean time to recovery.

In today's digital age, businesses rely on services and products to function seamlessly, with minimal downtime expected from these systems. Uptime is defined as the time when something is operational, but it can be measured in different ways, resulting in a spectrum of functionality. Measuring uptime is often done using nines, where each nine represents exponentially harder to achieve, such as 90% (one nine), 99% (two nines), and 99.9% (three nines) uptime. Four and five-nines require more validation and automation, and companies with these levels of uptime have a carefully cultivated engineering culture that encourages high quality and accountability. To improve uptime, developer teams can focus on error prevention and awareness, such as code validation and checks, automated testing, canary deploys, and application monitoring, which provide real-time insights into service health and alert developers to issues immediately. By implementing these strategies, businesses can minimize downtime and resolve issues quickly, ultimately improving service uptime.

Sentry's Integration Platform allows developers to build publicly available tools on top of Sentry, but also now offers a new feature called Internal Integrations that enables building internal tools with a simpler authentication model and access to Sentry's API. This new feature is designed to meet the needs of engineering teams who use proprietary or internal tools daily. With Internal Integrations, developers can easily integrate Sentry with their team's unique internal tools and workflow, and also leverage UI Augmentation features to enhance the integration experience. The platform provides a more streamlined way for developers to build custom tools that fit their specific needs, making it easier to work efficiently within their existing workflows.

Sentry has launched an Internal Integration Platform, allowing developers to build internal tools on top of Sentry for seamless integration into existing workflows. This platform simplifies authentication and offers a range of features such as Webhooks creation, accessing core API methods, and UI Augmentation. The new platform aims to improve efficiency by integrating with unique internal tools and workflows.

The Sentry unified Go SDK provides a comprehensive solution for error tracking, panics, breadcrumbs, contextual data, goroutines, and HTTP package integration. It supports Go Modules and integrates with various HTTP libraries such as net/HTTP, Echo, FastHTTP, Iris, Gin, Martini, Negroni, and others. The SDK offers features like recovering from panics, reporting errors, recording breadcrumbs, logging messages, extracting stack traces, filtering errors, and attaching contextual data to captured events. It provides methods for capturing exceptions, messages, and breadcrumbs, as well as configuring scope and attaching global data to events. The SDK also handles goroutines by utilizing the Hub and Clone functionality to ensure proper data communication between scopes and clients. With its extensive integration with HTTP packages, developers can easily capture panics and errors in their request handlers and expose the Request object for further disposal.

The Sentry unified Go SDK provides a comprehensive solution for error tracking, panic recovery, breadcrumbing, contextual data management, and integration with various HTTP libraries. The SDK is designed to utilize Go's features, such as recovering from panics, reporting errors, recording breadcrumbs, logging messages, extracting stack traces, filtering errors, and providing async/sync transports. It also supports serverless environments, thread-safe data separation, and extraction of request/os/device data. The SDK can be initialized with a DSN obtained from the Sentry account, and it provides methods for capturing exceptions, messages, breadcrumbs, and contextual data. Additionally, the SDK is designed to work seamlessly with goroutines, providing a Hub that keeps track of corresponding Scope instances and Client configurations. It also integrates with popular HTTP libraries such as net/HTTP, Echo, FastHTTP, Iris, Gin, Martini, Negroni, and others. The SDK's API is stable, but it is still in the "pre-v1" stage, and the development team welcomes feedback and requests for features to improve the solution.

Sentry's new Developer Advocate, Liz Krane, relies on various tools to manage her day-to-day tasks and maintain a work-life balance. She uses a combination of reading, writing, coding, and collaboration tools to research conferences, write talk proposals, develop code, and engage with the developer community. Liz also utilizes project management techniques such as Google Calendar and Todo lists to prioritize her tasks and allocate time for each activity. Additionally, she employs spaced repetition using Anki to commit new information to long-term memory and uses a custom-built Android app called Snooze Tracker to help her wake up on time. Despite not having strong preferences for coding tools or formatting styles, Liz values simplicity and convention over personal opinions.

Liz Krane, Sentry's new Developer Advocate, relies on a variety of tools and processes to manage her busy week, which includes researching conferences, writing talk proposals, coding, and hosting events. She uses a combination of online resources such as Google, YouTube, CFP Land, CallbackWomen, and GitHub to find relevant information and connect with others in the tech community. For writing and coding, Liz prefers Vim, using its Markdown syntax highlighting and command line tool Grip for formatting consistency. She also utilizes Glitch, a code sandbox, to experiment with full-stack applications and GitHub for documenting her personal projects. To stay organized, Liz uses Google Calendar, an open-source spaced repetition app Anki, and a custom Android app Snooze Tracker to manage her sleep schedule and wake up on time. Despite the importance of sleep in developer productivity, Liz believes that it is essential to find the right balance between work and rest.

This summer, two interns, Tony Xaio from the University of Toronto and Osmar Coronel from the University of Illinois, joined Sentry's bizops engineering team to work on projects supporting the Sentry for Data Teams initiative. They successfully integrated Sentry with Apache Airflow, enabling better instrumentation and metrics integration, as well as improving visibility into data pipelines. Furthermore, they tackled the challenge of integrating Sentry with Apache Beam, gaining insights into real-time data pipelines without parsing through logs in Stackdriver. The interns also participated in various after-hours activities, making a huge impact on Sentry during their summer internship.

This summer, two interns from the University of Toronto and the University of Illinois joined our bizops engineering team at Sentry, where they worked on several projects in support of the Sentry for Data Teams initiative. They successfully integrated Apache Airflow with Sentry, enabling better instrumentation and metrics integration with DataDog, as well as gaining visibility into their DAGs' execution. Additionally, they helped gain insights into Dataflow streaming pipelines using Apache Beam, which was a challenging but successful endeavor. The interns also participated in various after-hours activities, including attending an Airflow meetup and organizing an interns-only meetup. Their hard work had a huge impact on Sentry, and we wish them all the best as they head back to school.

Sentry is being used by Code for America's GetCalFresh initiative to monitor exceptions and errors in their digital application process, which aims to make government services more accessible and user-friendly. The team uses Sentry to gain visibility into errors, receive notifications of unexpected issues, and track error frequency. This allows them to quickly identify and debug problems, making improvements to the system and ultimately helping to close the participation gap for Californians who qualify for food assistance but are not receiving benefits. By leveraging Sentry's open-source nature, the team can also review source code and collaborate with other developers to solve implementation questions.

Sentry is being used by Code for America, a non-profit organization, to monitor exceptions in their GetCalFresh project. The team uses Sentry's error visibility and insights to identify and debug issues with the application process, which guides applicants through the enrollment process for food stamps. By using Sentry, the GetCalFresh team can quickly identify errors, track impact metrics, and gain insights into error frequency, ultimately helping them progress toward closing the participation gap in the CalFresh program.

Sentry has released new features for its SDKs, open source project, and other tools in June and July, including moving the Go SDK out of beta, enabling PDB file upload directly to Sentry, and making frontend development easier. The company also deprecated symbolserver in support of a standard symbolicator for iOS Symbolification, improved Freight deployment tool, and released Sentry 9.1.2. Additionally, Sentry has made changes to its admin interface, allowing admins to add and remove members from projects, and improving connection flows with Azure Devops and GitLab. The company is also working on a new code of conduct for its open source project.

Sentry has released several updates in June and July, including improvements to its SDKs for integrations with HTTP libraries, enabling direct upload of PDB files, and introducing reminders to update out-of-date SDKs. The company also made progress on its open-source project, adding a code of conduct, deprecating the symbolserver, and improving tools like Freight and frontend development. Additionally, Sentry enhanced its administrative features, allowing admins to manage project members and download usage reports, while also providing better feedback for Azure DevOps and GitLab connections.