Although the bug report is still open, there is a test already for it inside mysql-test/t/processlist_notembedded.test, and the test fails fairly frequently in buildbot:
10.2 5b93a483e41
main.processlist_notembedded w4 [ fail ]
Test ended at 2021-01-27 15:54:35
CURRENT_TEST: main.processlist_notembedded
mysqltest: At line 55: query 'SHOW EXPLAIN FOR $con_id' failed: 1933: Target is not running an EXPLAINable command
The result from queries just before the failure was:
< snip >
connect con1,localhost,root,,;
connection con1;
SET DEBUG_SYNC= 'before_join_optimize SIGNAL in_sync WAIT_FOR go';
connection default;
SET DEBUG_SYNC= 'now WAIT_FOR in_sync';
FOUND 1 /sleep/ in MDEV-20466.text
SET DEBUG_SYNC= 'now SIGNAL go';
connection con1;
user
disconnect con1;
connection default;
SET DEBUG_SYNC = 'RESET';
End of 5.5 tests
#
# MDEV-23752: SHOW EXPLAIN FOR thd waits for sleep
#
connect con1,localhost,root,,;
select sleep(100000);;
connection default;
SHOW EXPLAIN FOR con_id;
Please fix the test before you close the bug report.
Elena Stepanova
added a comment - Although the bug report is still open, there is a test already for it inside mysql-test/t/processlist_notembedded.test, and the test fails fairly frequently in buildbot:
10.2 5b93a483e41
main.processlist_notembedded w4 [ fail ]
Test ended at 2021-01-27 15:54:35
CURRENT_TEST: main.processlist_notembedded
mysqltest: At line 55: query 'SHOW EXPLAIN FOR $con_id' failed: 1933: Target is not running an EXPLAINable command
The result from queries just before the failure was:
< snip >
connect con1,localhost,root,,;
connection con1;
SET DEBUG_SYNC= 'before_join_optimize SIGNAL in_sync WAIT_FOR go';
connection default;
SET DEBUG_SYNC= 'now WAIT_FOR in_sync';
FOUND 1 /sleep/ in MDEV-20466.text
SET DEBUG_SYNC= 'now SIGNAL go';
connection con1;
user
disconnect con1;
connection default;
SET DEBUG_SYNC = 'RESET';
End of 5.5 tests
#
# MDEV-23752: SHOW EXPLAIN FOR thd waits for sleep
#
connect con1,localhost,root,,;
select sleep(100000);;
connection default;
SHOW EXPLAIN FOR con_id;
Please fix the test before you close the bug report.
ok, I see that at least one of my review requests are in the commit, but not all.
I also noticed the split of 'awake' is already done in 10.6, so a lot of the changes in the code for dec45aca77ee23557c59411fa9abf037e5343739 has already been done.
Can you please compare your patch against 10.6 and create a new one?
Also please test that the above test case really works with 10.6
Michael Widenius
added a comment - ok, I see that at least one of my review requests are in the commit, but not all.
I also noticed the split of 'awake' is already done in 10.6, so a lot of the changes in the code for dec45aca77ee23557c59411fa9abf037e5343739 has already been done.
Can you please compare your patch against 10.6 and create a new one?
Also please test that the above test case really works with 10.6
Bug
Critical
for 10.6:
commit dec45aca77ee23557c59411fa9abf037e5343739 (HEAD -> bb-10.6-MDEV-23752-2, origin/bb-10.6-MDEV-23752-2, origin/bb-10.6-MDEV-16440-2)Author: Oleksandr Byelkin <sanja@mariadb.com>Date: Fri Sep 18 17:35:08 2020 +0200MDEV-23752 SHOW EXPLAIN FOR thd waits for sleepmy_apc awake thread after request.for 10.2 & up (in 10.3 argument TRUE to check_killed() shoud be added)
commit 3d466b3c9b704e48bdffb9ed84852b9f3720b22b (HEAD -> bb-10.2-MDEV-23752, origin/bb-10.2-MDEV-23752)Author: Oleksandr Byelkin <sanja@mariadb.com>Date: Fri Sep 18 09:11:38 2020 +0200MDEV-23752 SHOW EXPLAIN FOR thd waits for sleepcall check_killed() periodically during sleep()