Details
Description
It just occurred to me that we are dropping 0ms spans when we poll the sendQueue for a span to deliver via thrift. It would cause less churn in the queue if we dropped the span as soon as we received it, before inserting it into the queue.
This is already fixed in 1.6.3 which no longer drops 0ms spans. It is only really a concern in 1.7.0 if you are using HDFS tracing, which generates a LOT of 0ms spans.
Attachments
Attachments
Issue Links
- is duplicated by
-
ACCUMULO-3863 Improve how AsyncSpanReceiver drops short spans
- Resolved