Looking Back on 2017

It’s still January, even if barely, which means this remains a completely appropriate time to post a 2017 year in review. 2015 was a big year for us. 2016 was even bigger. You know what that means: 2017 was the biggest yet.

How big is biggest?

Over half-a-million developers used Sentry last year, sending us right around 300 billion events in the process. Compare this to 2015, when we received 15 billion events. We’re all making (and resolving) a lot more errors together.

Our team more than doubled in size, growing from 20 employees at the end of 2016 to 53 at the end of 2017 (and 56 as of now). And to think, there were only two of us in 2015. We moved to a new office in San Francisco to accommodate all these new folks and expanded our office down under in Austria, the Australia of Europe.

As the team was still pretty small at the beginning of 2017, we didn’t have much of a defined process for getting work done. We essentially just did stuff until that stuff was finished. That worked, but we’re starting to get too big for that sort of process. So we got our shit together, created goal-oriented teams, adopted a sprint process, and hired our first PM. Here are the many things we got done (there’s so much, we now have to temporarily abandon paragraphs and dive straight into lists):

Q1

• Legacy browser filters
• Breadcrumbs for Node.js SDK
• Reprocessing for iOS crashes
• Official React Native support with split iOS/JavaScript traces

Q2

• Releases
  • including deploy emails, suggested owners, GitHub + Bitbucket integrations, Heroku deploy integration, and more
• Volume-based billing
• @Mentions in comments
• Java SDK 1.0
• Sentry in the GitHub Marketplace

Q3

• New resolution and silence workflow tools
• New project onboarding + platform-specific emails
• Similar issues, unmerging issues, and stack trace differential viewing
• Fine-grained client key rate limiting
• Invite multiple members at once

Q4

• Shared issue security redesign
• Enterprise SSO and SAML
• Discard and delete, additional inbound filters, and inbound filter stats
• Visual Studio Team Services integration
• Data Fordwarding
• Support for Minidump crash reports
• Automatic translation of IE error messages 😱

We also invested in our tools, our infrastructure, and our practices:

• Adopted Jest, Enzyme, and snapshots for JS unit tests
• Adopted Prettier and YAPF automatic code formatting
• Made getsentry/sentry configured out-of-the-box for use with Visual Studio Code
• Adopted React Storybook w/ Percy snapshots
• Migrated from Softlayer to Google Cloud ☁️ in four months
• Automated cross-browser testing for Raven.js (Sentry’s JavaScript client library)
• Added Snyk for automated JavaScript dependency testing
• Replaced LLVM with our own symbolic library written in Rust, greatly improving our symbolication performance
• Introduced our symbolicator service to “crowd source” the latest iOS debug symbols from users
• Adopted Veneur to to streamline the Sentry metric collection process
• Began using Redis Cluster in production
• Created Eos and other bots to automate our GitHub workflows
• Upgraded libraries, frameworks, and tools countless times

All this, on top of countless bug fixes, performance improvements, client library releases, and other things we’ve surely forgotten about.

Sentry engineers gave over 16 technical talks in over 9 different countries, including the US, France, UK, Germany, Italy, Belarus, Czech Republic, Russia, and Austria. For the first time, we gave talks at not just conferences and meetups, but also universities (Harvard) and code academies (Hack Reactor, App Academy).

We hosted 2 hack weeks this year, up from 1 last year. Some hack week projects either became or are on their way to becoming live Sentry features, like Relay and @Mentions.

So instead we’ll continue focusing on shipping great software with the goal of helping you ship even greater software. Sound like something you’d be interested in working on? We’d love your help!