You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Right now the error tagging doesn't support client side sampling, that is, if a trace is sampled out by the head of the trace, this information is not propagated in the trace header (aka dsc). This means that from the perspective of dynamic sampling, we don't know whether the trace has been sampled out on the client and this could lead to false positives.
We would need to fix this by having the client SDKs add the client sampling decision in the dsc that will be attached to the error. This client sampling decision will work similarly to dynamic sampling, that is, if the head is sampled out all the children are sampled out and vice versa.
The text was updated successfully, but these errors were encountered:
Had a meeting with @sl0thentr0py@smeubank@cleptric@jan-auer and we agreed that the SDKs team will drive the effort to add the sampled field in the DSC and Relay will take that in consideration when performing tagging.
Right now the error tagging doesn't support client side sampling, that is, if a trace is sampled out by the head of the trace, this information is not propagated in the trace header (aka dsc). This means that from the perspective of dynamic sampling, we don't know whether the trace has been sampled out on the client and this could lead to false positives.
We would need to fix this by having the client SDKs add the client sampling decision in the dsc that will be attached to the error. This client sampling decision will work similarly to dynamic sampling, that is, if the head is sampled out all the children are sampled out and vice versa.
The text was updated successfully, but these errors were encountered: