Vendors like Dynatrace, New Relic do only include traces in their error monitoring if the root trace itself is marked as an error, nested traces having an error are not enough.
We could provide an option to mark root spans as errors, if a nested instrumentation records an exception. We could also consider doing that by default, as from a "user operation" perspective, if some RPC call during a user-initiated request fails, then from a user perspective the operation has failed.
Vendors like Dynatrace, New Relic do only include traces in their error monitoring if the root trace itself is marked as an error, nested traces having an error are not enough.
We could provide an option to mark root spans as errors, if a nested instrumentation records an exception. We could also consider doing that by default, as from a "user operation" perspective, if some RPC call during a user-initiated request fails, then from a user perspective the operation has failed.