I may be wrong here, but I'd like to raise this to have a broader discussion.
On the OTel Demo we have a couple of services using flagd.evaluation.v*.Service/EventStream.
This creates a new span every X minutes.
- In Python we have spans that are 30 minutes long:
- In Java we have 10 minutes long spans:
- In Javascript we have 10 minutes long spans, all on the same trace:
The main issue to me is the span status ERROR.
If this connection is expected to end and reconnect, I don't think the span should be marked as error.
I may be wrong here, but I'd like to raise this to have a broader discussion.
On the OTel Demo we have a couple of services using
flagd.evaluation.v*.Service/EventStream.This creates a new span every X minutes.
EventStream flagd.evaluation.v2.Servicenot clearing trace context js-sdk-contrib#1522The main issue to me is the span status
ERROR.If this connection is expected to end and reconnect, I don't think the span should be marked as error.