Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
M4
-
None
-
None
Description
In a profile of a big scan, I see the TRACE call in the inner loop of HandleContinueScanRequest is costing a lot of CPU. This is likely because my particular scan is hitting MRS a lot, so each batch is only a few rows. But we should probably remove this trace or otherwise rate-limit it.