How does Sentry choose name of issue with multiple events with same fingerprint?

I use Symfony (php framework) and built-in Monolog logger.

Production usually configured with fingers_crossed type of logging, so all logs records will be stored only if there was one with error level. And this batch of records are related to particular request, so I set uuid of request as fingerprint of all records which will be sent as events to Sentry. Sentry shows only one issue that has all events in Related events so I have a trace of events to find out a problem. This solution works fine.

But I have a 2 problems in this case:

  1. Related events sorted in different order (not like logged errors in application) and have a title like Aug 17, 2017 7:16:33 AM UTC but not [app] emergency 1 message as error message.
    This problem is not a deal breaker, but it’s harder to read a list of events with same title and different culprit (or what that smaller subtitle called?)

  2. An issue has the same title and level as last event, but in my case it should be as event that caused an issue or at least as event with max error level
    With current behavior I can’t understand really reason of an issue in list view. I don’t see error level (because last event is usually has a debug level) and title of an issue (only debug message from last event). Also these issues with debug level are shown in filtered by level=error lists but without a label error.

a) Did I misunderstand a design of Sentry and related events particularly?
b) Maybe I didn’t find interface config for this. Does it exist?

1 Like