Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
11.2
-
None
Description
The reproducer may be non-deterministic, despite the debug sync point. On my machine the result depends on the number of rows in the table, below some row count the stages are shown as 3/4. I don't know whether it's related to the number of binlog events, or to their total size, or it is just a race condition of some sort. In the latter case you may be getting a different result on your machine/build, try to change (especially increase) the number of rows in the table.
Please don't put this test case into the regression suite, create a deterministic one instead.
--source include/have_innodb.inc
|
--source include/have_debug_sync.inc
|
--source include/have_sequence.inc
|
|
|
create table t (pk int, a int) engine=InnoDB; |
insert into t select seq, seq from seq_1_to_5000; |
|
|
--send set debug_sync= 'now wait_for go_dml'
|
|
|
--connect (con1,localhost,root,,)
|
set debug_sync= 'alter_table_online_before_lock signal go_dml wait_for go_alter'; |
--send alter table t force, algorithm=copy, lock=none;
|
|
|
--connection default
|
--reap
|
update t set a = a + 1; |
set debug_sync= 'now signal go_alter'; |
|
|
--let $run=100
|
while ($run)
|
{
|
query_vertical select stage, max_stage, info, state from information_schema.processlist where info like 'alter%'; |
--sleep 1 |
--dec $run |
if (`select count(*) = 0 from information_schema.processlist where info like 'alter%'`) |
{
|
--let $run= 0 |
}
|
}
|
|
|
|
|
--connection con1
|
--reap
|
drop table t; |
set debug_sync= reset; |
|
bb-11.2-oalter c29ff60b2c8 |
connection default; |
update t set a = a + 1; |
set debug_sync= 'now signal go_alter'; |
select stage, max_stage, info, state from information_schema.processlist where info like 'alter%'; |
stage 0
|
max_stage 0
|
info alter table t force, algorithm=copy, lock=none |
state Update_rows_log_event::ha_update_row(-1) on table `t` |
select stage, max_stage, info, state from information_schema.processlist where info like 'alter%'; |
stage 0
|
max_stage 0
|
info alter table t force, algorithm=copy, lock=none |
state Update_rows_log_event::find_row(-1) on table `t` |
select stage, max_stage, info, state from information_schema.processlist where info like 'alter%'; |
stage 0
|
max_stage 0
|
...
|
Attachments
Issue Links
- is caused by
-
MDEV-16329 Engine-independent online ALTER TABLE
-
- Closed
-
-
MDEV-28808
Test MDEV-16329 (ALTER ONLINE TABLE) - Core server part
-
- Closed
-