Sentry as a way not to be afraid

Irina Grechikhina

Debugging Development HTTP Web

See in schedule

Sentry is an open-source error tracking software. It allows you to know exactly what's going on with your web application in real-time and get detailed reports about occurred errors. Lot's of web developers used to rely only on app logs and generic monitoring software, but it's not enough. Multiple production servers, overwhelmed logging, lack of information about request content in case of error hinders problem investigation and produces permanent developer fear of changes. Sentry could ease the whole process and make you: stay calm during deployment, be sure your application works properly, solve problems before users report about it.

During my talk, I will compare Sentry to ordinary logging, show how to use Sentry, explain how to read Sentry reports and demonstrate that its deployment is very easy. The whole demo will be done using Sentry hosted on my laptop. I will describe the problems I have faced up during sentry.io usage and explain why I prefer self-hosting. Also, I will share a link to my Github with docker-compose and ansible files to deploy your self-hosted Sentry, run through the whole idea and demonstrate that everything is clear and easy. Then I will show how to adjust the first project in Sentry and connect it to a working application in both backend and frontend parts. For each backend and frontend cases, there will be its separate part with examples when Sentry is useful and cannot be replaced by other tools and recommendations on how to make Sentry as effective as possible.

My talk will be useful for web developers of any level who wants to ease the process of problem investigation and finds application logs not handy.

Type: Talk (45 mins); Python level: Beginner; Domain level: Beginner


Irina Grechikhina

CloudLinux

I wrote the first line of Python code 10 years ago. Began as a manual tester, became a QA team lead over time, and finally moved to backend Python development. I love doing educational projects and bringing knowledge (made an algorithm and data structure course for beginners, running a Telegram channel with tips and tricks on Python and Linux). I'm talking about Sentry every time I have the opportunity (PyCon KG 2018, PyCon ZA 2019, IWD Moscow 2020), because, without monitoring, you can't build a reliable web application. Love it when everything is automated, covered with tests and metrics, and holds high loads. I know the difference between 150 ms and 50 ms, can count the load, optimize and make the backend in Python fast, reliable and requiring a minimum amount of resources.