Percona Server moved to https://jira.percona.com/projects/PS

  1. Bug #1671152
  2. Comment #8

Comment 8 for bug 1671152

Revision history for this message
George Ormond Lorch III (gl-az) wrote :

This is strange Rick. With the current 'fix' I used exactly your table structure to create one TokuDB table and one InnoDB table. I then took your lua script and 'bash'ed it and got about 150K rows spread across 9 or 10 partitions into the TokuDB table. I then did an INSERT INTO test.BIG_STORAGE_INNO SELECT * FROM BIG_STORAGE_TOKU followed by ANALYZE TABLE on both tables.

So we now have two tables that are identical in content and partition distribution.

Attached are the SHOW INDEXES and EXPLAIN results from the two tables. They seem to be very close in responses and choose the same plans.

So I am unsure of where the differences lie. Maybe I'll see if I can get more rows in the tables for a larger sampling set.