fix: silent ignore of 4xx errors from reporting#5253
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #5253 +/- ##
==========================================
+ Coverage 73.20% 73.23% +0.02%
==========================================
Files 424 424
Lines 59967 59964 -3
==========================================
+ Hits 43898 43912 +14
+ Misses 13606 13595 -11
+ Partials 2463 2457 -6 ☔ View full report in Codecov by Sentry. |
|
Wouldn't it cause noise if we have a developer doing local testing with app.dev and prod reporting? Wouldn't it be better to classify the errors on reporting side and then add retries over here? |
We already have some error classification in place for the reporting service, and we retry these errors as needed. However, consider a recent issue we encountered: if the source definition ID is incorrect, the reporting service responds with a 400 error. As a result, those reports are completely lost. |
…rrors-in-reporting-service
Description
We are currently ignoring 4XX errors from reporting service (for example due to payload too large or invalid sourceDefinitionId value). This result in reports being silently dropped without generating any alert.
By treating 4XX as failure, we ensure that reports are retired, allowing us to detect and receive alerts for high 4XX rates or report backlogs.
Linear Ticket
https://linear.app/rudderstack/issue/OBS-711/fix-silent-ignore-of-400-errors-in-reporting-service
Security