November 2017 Summaries

The sentry-cli command line client can now send events to Sentry, making automatic error reporting from bash scripts possible. To do this, users need a working sentry-cli installation and must configure it with their DSN for events. With the -m parameter and a message for the event, users can start sending basic events. Additionally, they can attach breadcrumbs to events by providing logfiles or other relevant information. By adding one line at the start of bash scripts, users can enable error handling and receive tracebacks and output as breadcrumbs after a failure in a script.

The text discusses managing errors with Sentry's Inbound Filters. It explains that some errors may be ignored due to various reasons such as being minor or already resolved. Sentry provides inbound filters for different subscription levels, allowing users to determine which errors should be dropped. These filters include filtering by error message, release, IP address, web crawlers, legacy browsers, browser extensions, and localhost. Using these filters can enhance the user experience with Sentry and make it a more valuable service.

Sentry has expanded its Talent team with the addition of Maggie Bauer, who joins from Twitch where she helped quadruple the number of employees in 2 1/2 years by improving recruiting processes and optimizing talent operations.

In this article, Divydeep Agarwal, CTO of Branch Messenger, shares a story about a bug that prevented certain users from logging into their accounts after the launch of a new web build. The issue was found to be related to Angular's built-in date pipe and only affected Internet Explorer (IE) users. With the help of error tracking system Sentry, the team discovered the specific compatibility issues with IE and fixed the problem in about four hours. This highlights the importance of using an error tracking system to quickly identify and fix bugs that impact user experience.

Chris Clark joins the Sentry Team, bringing design and frontend expertise from the Genius team in NYC, where he spent several years engaging in debates about Eminem vs Jay Z, to contribute to various projects at Sentry, while also pursuing his hobbies of rock climbing, coffee appreciation, and creating an online cocktail database.

The Sentry team participated in their winter HACKWEEK, a week-long event where employees focused on various projects that might not have been completed otherwise. Nearly all employees attended, with some coming from far-flung locales, and worked on projects ranging from improvements to Sentry itself to enhancing office experiences. Fourteen projects were presented, with six winning awards, including the Best Overall Project, Strut, which brings entrance theme music to employees when they enter the office. Other notable projects included Value People: Get Lit, a project that uses LED boards to flash employee names, and Care Hard: Project SDK / Setup Wizard, an update to Sentry's setup process. The team also worked on making their internal knowledge base searchable and created various apps, including one that turns an iPhone X into a shock-free taser.

The author has been working with Sentry's Ruby client for a couple of years, initially switching from a non-Sentry monitoring service due to poor performance and user experience. They've made significant contributions to the client and now work as a paid contributor. The two key features that improve error tracking are transactions and async processing. Transactions provide detailed information about where an error occurred in the application's workflow, allowing developers to quickly identify issues and resolve them. This is achieved through the Transaction API, which can be used from the moment the Ruby gem is installed. Async processing allows for asynchronous sending of exceptions to Sentry, enabling faster response times without blocking the main thread.

We use Milksnake, a Python module that hooks into the setuptools system with CFFI to execute external build commands, to build native shared libraries which then in turn get loaded by Python through CFFI. This allows us to build Rust code as Python packages without requiring a Rust compiler on Mac and Linux platforms. The use of Milksnake enables us to generate very specific type of Python wheel that is independent of the Python version or implementation, allowing for greater flexibility and compatibility across different platforms. By using this approach, we can expose our C-ABIs in a way that makes sense in C, which simplifies the integration with other languages such as C and C++. The system uses cbindgen to generate C headers from our Rust code, and then generates high-level Python wrappers around these low-level bindings. It also includes automatic error handling and panic handling through macros, making it easier to work with the generated C-ABIs. The system is designed to anchor the lifetime of objects in the Python space, invoking deallocation functions on the Rust/C side. This allows for more control over memory management and makes the integration with other languages more seamless.

Maya Porter joins Sentry as an Office Operations Manager, bringing experience from a Co-op Intern at VMware, Inc., where she developed a spotlight series to help employees shine, and her studies in marketing from Northeastern University. In this role, she will oversee day-to-day processes and aim to create a positive work environment. Maya is also a self-proclaimed psychology nerd with interests outside of work, including working with kids, eating açaí, and drinking coffee.

The Sentry Elixir SDK has come a long way since its inception, with significant features and configuration flexibility added over the past year. The introduction of Sentry.Plug allows for easy error reporting in Plug-based applications, while Sentry.Logger provides a solution for tracking down issues in complex supervision trees and applications. One unique feature is Sentry.Sources, which enables viewing source code alongside an error report on Sentry's website. The SDK has become a full-featured client with the same features as its popular Ruby and JavaScript siblings, thanks to Sentry's support. The maintainer credits Jason Stiebs for his significant contributions and help along the way.

The conversation with Valentino Volonghi, CTO of AdRoll, revolves around an intriguing application and infrastructure issue they encountered in the past. The company was experiencing a critical failure without any machine crashes or system malfunctions, which led to the adoption of Amazon's new C3 Instance Family for improved performance. However, after the upgrade, the machines weren't spending their budgets efficiently, causing a drop in revenue due to reduced processing power. An investigation revealed that long-lived connections were being held by specific instances, leading to an imbalance in machine utilization and budget allocation. To resolve this issue, AdRoll introduced random timeouts into the software to encourage rebalancing of connection allocation across different machines, effectively allocating resources efficiently even for long-lasting connections.

Chloe Condon is a former musical theatre actress who has transitioned into a role at Sentry, a company that aims to bring people with non-traditional backgrounds into the world of tech. Chloe is passionate about this mission and can be seen at various Sentry events, speaking at conferences, or engaging in other activities such as crafting and blogging. She has a diverse background, having been involved in musical theatre productions like "Hairspray", "Xanadu", and "Jerry Springer: the Opera". Chloe's enthusiasm for tech is reflected in her efforts to create content around Docker and other topics, showcasing her versatility and dedication to sharing knowledge with others.

Sentry recently updated its onboarding flow to collect additional data about the platforms users are coding on, with the goal of improving integration and user experience. The new platform data has revealed insights into how different languages and integrations perform in terms of ease of use and time taken to send a first event. Python and PHP projects have emerged as the most successful, with high success rates and short times to reach a first event, while JavaScript and Ruby also show promising results, with users sending their first event within two hours. This new data will inform engineering improvements aimed at simplifying integration with Sentry for all users.