[MDEV-5497] Test suite test connect.fix fails Created: 2014-01-01  Updated: 2014-06-16  Resolved: 2014-06-16

Status: Closed
Project: MariaDB Server
Component/s: None
Affects Version/s: 10.0.7, 10.0.8
Fix Version/s: 10.0.12

Type: Bug Priority: Minor
Reporter: Duncan Roe Assignee: Alexander Barkov
Resolution: Fixed Votes: 0
Labels: None
Environment:

Slackware Linux 14.1 + kernel 3.12.6


 Description    

cd /usr/mysql-test; ./mtr connect.fix

(as root) gives me an error report as below. I am keeping the mysql-test directory, so please let me know if you'd like anything else from it.

Logging: ./mtr  connect.fix
 
vardir: /usr/mysql-test/var
 
Checking leftover processes...
 
Removing old var directory...
 
Creating var directory '/usr/mysql-test/var'...
 
Checking supported features...
 
MariaDB Version 10.0.7-MariaDB
 
 - SSL connections supported
 
Collecting tests...
 
Installing system database...
 
Using server port 43317
 
 
 
==============================================================================
 
 
 
TEST                                      RESULT   TIME (ms) or COMMENT
 
--------------------------------------------------------------------------
 
 
 
worker[1] Using MTR_BUILD_THREAD 300, with reserved ports 16000..16019
 
worker[1] mysql-test-run: WARNING: running this script as _root_ will cause some tests to be skipped
 
connect.fix                              [ fail ]
 
        Test ended at 2014-01-01 19:32:13
 
 
 
CURRENT_TEST: connect.fix
 
--- /usr/mysql-test/plugin/connect/connect/r/fix.result 2013-12-23 20:35:03.000000000 +1100
 
+++ /usr/mysql-test/plugin/connect/connect/r/fix.reject 2014-01-01 19:32:13.000000000 +1100
 
@@ -80,7 +80,7 @@
 
 name   city    birth   hired
 
 John   Boston     2    1986-01-05      2010-06-02
 
 Henry  Boston     0    1987-06-07      2008-04-01
 
-George San Jose   1    1981-08-01      2010-06-02
 
+George San Jose   1    1981-07-31      2010-06-02
 
 Sam    Chicago    2    1979-11-02      2007-10-10
 
 James  Dallas     1    1992-05-03      2009-12-14
 
 Bill   Boston     1    1986-09-01      2008-02-10
 
@@ -96,7 +96,7 @@
 
 name   city    birth   hired
 
 John   Boston     2    1986-01-05      2010-06-02
 
 Henry  Boston     0    1987-06-07      2008-04-01
 
-George San Jose   1    1981-08-01      2010-06-02
 
+George San Jose   1    1981-07-31      2010-06-02
 
 Sam    Chicago    2    1979-11-02      2007-10-10
 
 James  Dallas     1    1992-05-03      2009-12-14
 
 Bill   Boston     1    1986-09-01      2008-02-10
 
@@ -112,7 +112,7 @@
 
 name   city    birth   hired
 
 John   Boston     2    1986-01-05      2010-06-02
 
 Henry  Boston     0    1987-06-07      2008-04-01
 
-George San Jose   1    1981-08-01      2010-06-02
 
+George San Jose   1    1981-07-31      2010-06-02
 
 Sam    Chicago    2    1979-11-02      2007-10-10
 
 James  Dallas     1    1992-05-03      2009-12-14
 
 Bill   Boston     1    1986-09-01      2008-02-10
 
@@ -128,7 +128,7 @@
 
 name   city    birth   hired
 
 John   Boston     2    1986-01-05      2010-06-02
 
 Henry  Boston     0    1987-06-07      2008-04-01
 
-George San Jose   1    1981-08-01      2010-06-02
 
+George San Jose   1    1981-07-31      2010-06-02
 
 Sam    Chicago    2    1979-11-02      2007-10-10
 
 James  Dallas     1    1992-05-03      2009-12-14
 
 Bill   Boston     1    1986-09-01      2008-02-10
 
 
 
mysqltest: Result content mismatch
 
 
 
 - saving '/usr/mysql-test/var/log/connect.fix/' to '/usr/mysql-test/var/log/connect.fix/'
 
--------------------------------------------------------------------------
 
The servers were restarted 0 times
 
Spent 0.000 of 6 seconds executing testcases
 
 
 
Failure: Failed 1/1 tests, 0.00% were successful.
 
 
 
Failing test(s): connect.fix
 
 
 
The log files in var/log may give you some hint of what went wrong.
 
 
 
If you want to report this error, please read first the documentation
 
at http://dev.mysql.com/doc/mysql/en/mysql-test-suite.html
 
 
 
mysql-test-run: *** ERROR: there were failing test cases



 Comments   
Comment by Sergei Golubchik [ 2014-01-19 ]

bar, looks like a timezone issue, doesn't it?

Comment by Duncan Roe [ 2014-01-27 ]

Timezone here is AEDT (GMT + 11:00). However, the database is loaded and read with the same timezone, so how would it make a difference?

Comment by Alexander Barkov [ 2014-01-29 ]

The problem is reproducible on a Fedora 18 box
with time zone set to EST (Australia/Sydney).

Comment by Duncan Roe [ 2014-02-11 ]

Still failing at 10.0.8

Comment by Elena Stepanova [ 2014-03-04 ]

The result file looks totally weird to me:
e.g.

 name   city    birth   hired
 
John   Boston     2    1986-01-05      2010-06-02

Which column "2" belongs to?
The data file only has 4 fields, and the table has 4 columns, so "2" is attached either to city or to birth.
The intial data has "25/01/1986" as birth date for John Boston, but it looks like the first digit is taken apart and treated separately, so the date only has "5/01/1986", which is converted into 1986-01-05.
I guess this conversion is not very deterministic for "0/08/1981", and the test fails with some timezones

Comment by Otto Kekäläinen [ 2014-03-04 ]

I have reproduced this issue building MariaDB 10.0 on a machine with the timezone UTC+2.
Temporairly disabled test in https://github.com/ottok/mariadb-10.0/commit/2c9b7451

Comment by Olivier Bertrand [ 2014-03-05 ]

Apparently non-sense result comes from wrong create table statements.
In the boys data file, the field "city" is 11 characters long, not 12. Therefore the first digit of the birth date was added to the city field (after some blanks)

Also, this character was removed from the birth date. In the case of the of George in San Jose, the date became 0/08/1981 giving a result that seems to depend on the time zone. Is this a bug?

Comment by Olivier Bertrand [ 2014-03-05 ]

Sorry Elena, I did not read your comment entirely before answering. Your analyze was quite good, just missing the reason why this was happening.

Comment by Sergei Golubchik [ 2014-03-05 ]

Reopened. Even if the behavior is expected and there's no bug in the CONNECT engine, the test still fails. it's not good, the test needs to be fixed to be deterministic.

Comment by Olivier Bertrand [ 2014-03-05 ]

As a matter of facts, after I closed it, I forgot to push my fix.
I just did it now, and you can test if it really fixed it in different
time zones.

Generated at Thu Feb 08 07:04:50 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.