Bug #28350 Different number of rows reported by "explain"

• Submitted: 10 May 2007 11:29
• Modified: 22 Jun 2007 11:44
• Reporter: Joerg Bruehe
• Status: Can't repeat
• Category: MySQL Server: Row Based Replication ( RBR )
• Severity: S2 (Serious)
• Version: 5.1.18
• OS: Linux (x86_64)
• Assigned to: Alexey Botchkov
• CPU Architecture: Any

[10 May 2007 11:29] Joerg Bruehe

Description:
New effect in 5.1.18-beta.

Occurred only in a RPM build on Linux/x86_64,
in both tests using row-based replication:
"normal+rowrepl"  and  "ps+rowrepl+NDB"

-------------------------------------------------------
*** r/innodb_mysql.result
--- r/innodb_mysql.reject
***************
*** 380,386
  DELETE FROM t1;
  EXPLAIN SELECT DISTINCT t1.name, t1.dept FROM t1 WHERE t1.name='rs5';
  id    select_type     table   type    possible_keys   key     key_len ref     rows    Extra
! 1     SIMPLE  t1      range   name    name    44      NULL    2       Using where; Using index for group-by
  SELECT DISTINCT t1.name, t1.dept FROM t1 WHERE t1.name='rs5';
  name  dept
  DROP TABLE t1;
--- 380,386
  DELETE FROM t1;
  EXPLAIN SELECT DISTINCT t1.name, t1.dept FROM t1 WHERE t1.name='rs5';
  id    select_type     table   type    possible_keys   key     key_len ref     rows    Extra
! 1     SIMPLE  t1      range   name    name    44      NULL    1       Using where; Using index for group-by
  SELECT DISTINCT t1.name, t1.dept FROM t1 WHERE t1.name='rs5';
  name  dept
  DROP TABLE t1;
-------------------------------------------------------

How to repeat:
Occurred when running the test suite in the release build.

[22 Jun 2007 11:44] Joerg Bruehe

Seems to be a spurious failure -
it did not happen again when building 5.1.19-beta.

There, the test passed in all runs
except for bug 23810 (which is a different thing: Solaris-Sparc-64, debug).

Setting to "Can't repeat", to be changed if we ever meet it again.