Improving your Sentry Debugging Experience with Insights into Feature Flags

[jas-kas]

📄 Background

We, at Sentry, use feature flags regularly to change our application behaviour and appearance conveniently without deploying new code. For example, our development process includes slowly rolling out new features to ensure they don’t introduce new errors and collect early product feedback.

We know that feature flag changes are just as valuable and critical to the debugging process as are code releases for some organizations. As such, we're looking at enhancing our context to include insights on your feature flags and integrating with popular feature flag systems. For example, displaying flags evaluated at the time of an error event, in order, on our Issue Details page.

👋 We’d like to hear from you

• How important are feature flags in your development process and how do you use them?
• What feature flagging system are you using today?
• When using Sentry to debug issues, what information would you want from your feature flag system there and how would you want it presented?
• What do you monitor to know if it is safe to continue to rollout a feature flag, and how can Sentry help facilitate this process?
• What kind of issues have you encountered that were related to feature flag changes and how did you go about debugging them today? How often do those types of issues happen?

Please share your initial thoughts and impressions below. 👇 If you’re interested in chatting live with the Sentry team on this topic, let us know below and we can schedule a call. Thank you!

[volokluev]

When using Sentry to debug issues, what information would you want from Flagpole [or insert other feature flag system] there?

Snuba has its own runtime config that reads from redis. Knowing the feature flags that were toggled at what time could be good for sure

How do you monitor if it's safe to continue to rollout a feature flag?

At this time I am switching between datadog for understanding infra load and sentry for business logic issues

What kind of issues have you encountered that were related to feature flag changes and how did you go about debugging them today?

Often, there's something in the sentry repo that is changed (by turning on a feature flag) that will cause a lot of downstream impacts on the infra. At the moment the way I debug them is go into #team-engineering and ask "yo what's going on, who broke this"