Description:
PB on mysql-5.1-bugteam :
main.log_tables [ fail ]
Test ended at 2009-02-02 12:34:18
CURRENT_TEST: main.log_tables
--- /data0/pushbuild/pb/bzr_mysql-5.1-bugteam/426/mysql-5.1.32-pb426/mysql-test/r/log_tables.result 2009-02-02 13:34:11.000000000 +0300
+++ /data0/pushbuild/pb/bzr_mysql-5.1-bugteam/426/mysql-5.1.32-pb426/mysql-test/r/log_tables.reject 2009-02-02 14:34:18.000000000 +0300
@@ -225,6 +225,7 @@
TIMESTAMP USER_HOST THREAD_ID 1 Query truncate table mysql.slow_log
TIMESTAMP USER_HOST THREAD_ID 1 Query set session long_query_time=1
TIMESTAMP USER_HOST THREAD_ID 1 Query select sleep(2)
+TIMESTAMP USER_HOST THREAD_ID 1 Quit
TIMESTAMP USER_HOST THREAD_ID 1 Query select * from mysql.slow_log
TIMESTAMP USER_HOST THREAD_ID 1 Query set @@session.long_query_time = @saved_long_query_time
TIMESTAMP USER_HOST THREAD_ID 1 Query alter table mysql.general_log engine=myisam
mysqltest: Result content mismatch
It looks like the test met a session belonging
to the previous test (There is no
"disconnect <session>" within the current test
before our current statement) where the "removal"
from the server till it is no more visible within
processlist needed too much time.
- Such effects are typical for a testing box with
heavy parallel load
- main.log_tables itself is not "guilty"
Preceding tests must take care that all
additional sessions are
- get a "disconnect <session>"
- disappeared from processlist when the test ends
Bug#42003 tests missing the disconnect of connections <> default
should take care of such "non cooperative" tests.
Therefore I set this bug to "Duplicate".
How to repeat:
See PB results
Description: PB on mysql-5.1-bugteam : main.log_tables [ fail ] Test ended at 2009-02-02 12:34:18 CURRENT_TEST: main.log_tables --- /data0/pushbuild/pb/bzr_mysql-5.1-bugteam/426/mysql-5.1.32-pb426/mysql-test/r/log_tables.result 2009-02-02 13:34:11.000000000 +0300 +++ /data0/pushbuild/pb/bzr_mysql-5.1-bugteam/426/mysql-5.1.32-pb426/mysql-test/r/log_tables.reject 2009-02-02 14:34:18.000000000 +0300 @@ -225,6 +225,7 @@ TIMESTAMP USER_HOST THREAD_ID 1 Query truncate table mysql.slow_log TIMESTAMP USER_HOST THREAD_ID 1 Query set session long_query_time=1 TIMESTAMP USER_HOST THREAD_ID 1 Query select sleep(2) +TIMESTAMP USER_HOST THREAD_ID 1 Quit TIMESTAMP USER_HOST THREAD_ID 1 Query select * from mysql.slow_log TIMESTAMP USER_HOST THREAD_ID 1 Query set @@session.long_query_time = @saved_long_query_time TIMESTAMP USER_HOST THREAD_ID 1 Query alter table mysql.general_log engine=myisam mysqltest: Result content mismatch It looks like the test met a session belonging to the previous test (There is no "disconnect <session>" within the current test before our current statement) where the "removal" from the server till it is no more visible within processlist needed too much time. - Such effects are typical for a testing box with heavy parallel load - main.log_tables itself is not "guilty" Preceding tests must take care that all additional sessions are - get a "disconnect <session>" - disappeared from processlist when the test ends Bug#42003 tests missing the disconnect of connections <> default should take care of such "non cooperative" tests. Therefore I set this bug to "Duplicate". How to repeat: See PB results