Detection of sources in analytics_sources.pipe is done using the current domain from the first event that arrives. This works fine on setups where there is only a single source, but when there are many it can lead to false positives: the current domain may change from event to event.
Expected Behavior
When multiple events arrive together they should be processed using the current domain from each one.
To Reproduce
Send multiple events with different href and referrer. The current domain from the first event is used for all of them.
Details
This behavior is particularly problematic with the Tinybird gatherer, as it coalesces multiple events and sends them together. It also leads to memory issues in certain setups with billions of events.
Module
Tinybird Data Project
Describe the Bug
analytics_sources.pipeis done using the current domain from the first event that arrives. This works fine on setups where there is only a single source, but when there are many it can lead to false positives: the current domain may change from event to event.Expected Behavior
When multiple events arrive together they should be processed using the current domain from each one.
To Reproduce
Send multiple events with different
hrefandreferrer. The current domain from the first event is used for all of them.Details
This behavior is particularly problematic with the Tinybird gatherer, as it coalesces multiple events and sends them together. It also leads to memory issues in certain setups with billions of events.