MySQL Bugs: #80582: Loose index scans not supported by ndbcluster engine?

Bug #80582	Loose index scans not supported by ndbcluster engine?
Submitted:	2 Mar 2016 9:59	Modified:	12 May 2016 5:18
Reporter:	Hartmut Holzgraefe	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S3 (Non-critical)
Version:	mysql-cluster-7.4.7	OS:	Linux
Assigned to:		CPU Architecture:	Any

Description:
On a table with a combined key on (col1, col2), with only 100 distinct values in col1, but a few 100K rows total, MyISAM and InnoDB can perform a "Loose Index Scan" on e.g.

  SELECT DISTINCT col1 FROM table;

which leads to only 100 rows being shown in the EXPLAIN plan, together with "Using index for group-by", and Handler_read_next goes up by ~100 when executing the actual query.

When doing this on MySQL Cluster EXPLAIN shows that a full table scan will be performed, and Handler_read_rnd_next goes up by the number of actual rows in the table.

I assume that the internal index operations required for a loose index scan would actually require even more network round trips than the full scan (as I don't remember any NDBAPI operations that would support this in an effective way) ...

How to repeat:
see above

Suggested fix:
Not sure whether this is an actual bug, a feature request, or just needs to be stated in the manual section on "loose index scans" that this might not be supported by all storage engines ...?

Hi Hartmut,

I filed it as documentation bug as you are right, this should be properly documented.

thanks for the submission

take care
Bogdan

If you don't know for certain whether it's a software issue or a docs issue, refer it to Dev. I've done so here.