Bug #32207 monitoring count of full table scans becomes useless with small memory tables
Submitted: 9 Nov 2007 2:22 Modified: 9 Nov 2007 11:16
Reporter: Dewey Gaedcke Email Updates:
Status: Verified Impact on me:
Category:MySQL Server: General Severity:S4 (Feature request)
Version:any OS:Any
Assigned to: CPU Architecture:Any
Tags: full table scan memory temporary

[9 Nov 2007 2:22] Dewey Gaedcke
For MySQL users who are using small, temporary, memory tables for 
advanced array-like tricks, the server is logging these selects as 
"full table scans".  While technically accurate, I don't think this 
implementation serves the main purpose behind this feature.  These 
types of queries should not be counted as it ruins the meaningfulness 
of this server statistic. 

How to repeat:
create temp memory table
populate it with 10 rows
run insert on update against a disk based table
you will have a log showing a full table scan

but it's meaningless because it was a 10 row memory table and has better performance than using indexes

Suggested fix:
 We should have a way to turn off monitoring of memory or temporary tables, or specific tables, or tables with < x rows.
[9 Nov 2007 11:16] Valeriy Kravchuk
Thank you for a reasonable feature request.