We have cases when we send an event with the same name shortly one after another (say, 5 seconds) then the consequent event is not recorded (but we see in logs that both messages have been sent to server). So it looks like some load filtering is taking place.
Is that some defined behaviour of the Sentry server? Can I configure it somehow to lower values?
Hi, if you send the event with the same (non-null/empty) value for event_id we will filter out the second event, as we consider it an exact duplicate based on that.
But we do not assign the event IDs manually, the Sentry SDK does this for us, right? And in our case, the event IDs are different: here just reproduced it again
2018-11-06 14:09:12
“event_id” : “CF19B759A8714782BF08566DAD3971DA”,
2018-11-06 14:09:28
“event_id” : "4C98545C2A454B20AE3EA7454BAB8D20”,
First is recorder (available on server), the second not - timestamp difference is 16 seconds.
In case someone has the same situation: we have figured this out my ourselves. It turned out Sentry groups those messages in one. Which is actually no something one should expect, because breadcrumb sets are different.
As a fix, we are adding some “fake” fingerprint to every message now