List:Commits« Previous MessageNext Message »
From:Øystein Grøvlen Date:December 23 2008 8:52am
Subject:Re: bzr commit into mysql-6.0-backup branch (hema:2744) WL#4229
View as plain text  
Hema, thanks, for addressing my latest comments.  We are getting there
now.  I will approve this now, but I ask you to fix one test failure I
observed and a potential instability.  I also hope that you make more
details commit comments, and also specific to each file, in your final
commit.

STATUS
======

Approved with 2 pending requests.


REQUESTS
========

1. I get a test failure if a run backup_logs after another test.  E.g,
    when I run the entire backup suite, I get:

       --- 
/export/home/tmp/oysteing/mysql/mysql-6.0-backup/mysql-test/suite/backup/r/backup_logs.result

       Tue Dec 23 11:09:58 2008
       +++ 
/export/home/tmp/oysteing/mysql/mysql-6.0-backup/mysql-test/suite/backup/r/backup_logs.reject

       Tue Dec 23 11:11:55 2008
       @@ -52,7 +52,7 @@
        SELECT MAX(backup_id) INTO @backup_id_history FROM 
mysql.backup_history
        WHERE command LIKE "BACKUP DATABASE backup_logs TO%";
        Verify the result
       -LET =`SELECT @backup_id_history = 270 AS are_identical`
       +LET =`SELECT @backup_id_history = 353 AS are_identical`

        Verification of backup_id from history table and command is:
        1

       mysqltest: Result content mismatch


    Note that when you echo your statements, do not include the
    variable.  Instead of:
       > +--echo LET $result=`SELECT @backup_id_history = $backup_id AS 
are_identical`
    you should do:
       > +--echo LET $result=`SELECT @backup_id_history = backup_id AS 
are_identical`
    or something like that.

2. I fear that the following might create infrequent failures:

       > +SELECT timediff(start_time, @start_backup) > 0 from 
mysql.backup_history
       > +WHERE backup_id=500;
       > +
       > +SELECT timediff(stop_time, @stop_backup) > 0 from 
mysql.backup_history
       > +WHERE backup_id=500;

    I do not think we can guarantee that start_time and @start_backup
    is always equal since even if the difference is just a few
    microsec, the clock could have increment it seconds part in the
    middle.  Hence, I think a number greater than 0 needs to be used.
    Note also that this assumes that you have control over which is
    recorded first of start_time and @start_backup.  If not, I guess
    you need to use an absolute value here.

SUGGESTIONS
===========

3. As an alternative to this:

       > +--echo Verify the result
       > +--echo LET $result=`SELECT @backup_id_history = $backup_id AS 
are_identical`
       > +LET $result=`SELECT @backup_id_history = $backup_id AS 
are_identical`;
       > +
       > +--echo
       > +--echo Verification of backup_id from history table and 
command is:
       > +--echo $result

    I suggest this:

       > +--echo Verify the backup id
       > +--disable query_log
       > +--echo SELECT @backup_id_history = backup_id AS are_identical
       > +--eval SELECT @backup_id_history = $backup_id AS are_identical
       > +--enable_query_log

    disable_query_log is used to in order to not expose $backup_id.
    The result of the query will still be printed.

--
Øystein

Hema Sridharan wrote:
 > #At file:///export/home/tmp/wl-4229/mysql-6.0-backup/ based on 
revid:charles.bell@stripped
 >
 >  2744 Hema Sridharan	2008-12-22
 >       WL#4229. New patch with some modifications.
 > added:
 >   mysql-test/include/blackhole.inc
 > modified:
 >   mysql-test/suite/backup/r/backup_errors.result
 >   mysql-test/suite/backup/r/backup_logs.result
 >   mysql-test/suite/backup/t/backup_errors.test
 >   mysql-test/suite/backup/t/backup_logs.test

...
Thread
bzr commit into mysql-6.0-backup branch (hema:2744) WL#4229Hema Sridharan22 Dec
  • Re: bzr commit into mysql-6.0-backup branch (hema:2744) WL#4229Øystein Grøvlen23 Dec