Commit 295ad743 authored by Annamalai Gurusami's avatar Annamalai Gurusami

Bug #14200010 NEWLY CREATED TABLE DOESN'T ALLOW FOR LOOSE INDEX SCANS

Problem:

Before the ALTER TABLE statement, the array
dict_index_t::stat_n_diff_key_vals had proper values calculated
and updated.  But after the ALTER TABLE statement, all the values
of this array is 0.  

Because of this statistics returned by innodb_rec_per_key() is
different before and after the ALTER TABLE statement. Running the
ANALYZE TABLE command populates the statistics correctly.

Solution:

After ALTER TABLE statement, set the flag dict_table_t::stat_initialized
correctly so that the table statistics will be recalculated properly when
the table is next loaded.  But note that we still don't choose the loose
index scans.  This fix only ensures that an ALTER TABLE does not change
the optimizer plan.

rb://1639 approved by Marko and Jimmy.
parent 3ed9ce93
...@@ -963,7 +963,7 @@ Table Op Msg_type Msg_text ...@@ -963,7 +963,7 @@ Table Op Msg_type Msg_text
test.t1 check status OK test.t1 check status OK
explain select * from t1 where b like 'adfd%'; explain select * from t1 where b like 'adfd%';
id select_type table type possible_keys key key_len ref rows Extra id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t1 ALL b NULL NULL NULL 15 Using where 1 SIMPLE t1 range b b 769 NULL 7 Using where
create table t2(a int, b varchar(255), primary key(a,b)) engine=innodb; create table t2(a int, b varchar(255), primary key(a,b)) engine=innodb;
insert into t2 select a,left(b,255) from t1; insert into t2 select a,left(b,255) from t1;
drop table t1; drop table t1;
......
...@@ -343,7 +343,7 @@ id select_type table type possible_keys key key_len ref rows Extra ...@@ -343,7 +343,7 @@ id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t1 index NULL PRIMARY 5 NULL 4 Using index; Using temporary 1 SIMPLE t1 index NULL PRIMARY 5 NULL 4 Using index; Using temporary
explain select distinct f1, f2 from t1; explain select distinct f1, f2 from t1;
id select_type table type possible_keys key key_len ref rows Extra id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t1 range NULL PRIMARY 5 NULL 3 Using index for group-by; Using temporary 1 SIMPLE t1 index NULL PRIMARY 5 NULL 4 Using index
drop table t1; drop table t1;
CREATE TABLE t1 (id int(11) NOT NULL PRIMARY KEY, name varchar(20), CREATE TABLE t1 (id int(11) NOT NULL PRIMARY KEY, name varchar(20),
INDEX (name)); INDEX (name));
......
...@@ -4083,6 +4083,13 @@ end: ...@@ -4083,6 +4083,13 @@ end:
trx->error_state = DB_SUCCESS; trx->error_state = DB_SUCCESS;
trx_general_rollback_for_mysql(trx, NULL); trx_general_rollback_for_mysql(trx, NULL);
trx->error_state = DB_SUCCESS; trx->error_state = DB_SUCCESS;
} else {
if (old_is_tmp && !new_is_tmp) {
/* After ALTER TABLE the table statistics
needs to be rebuilt. It will be rebuilt
when the table is loaded again. */
table->stat_initialized = FALSE;
}
} }
} }
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment