List:Commits« Previous MessageNext Message »
From:Magnus Svensson Date:October 16 2007 7:52am
Subject:Re: bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099
View as plain text  
Hi Georgi,

I don't really see how this patch is related to the problem at hand. The
original problem the bug reports is that you get and additional warning
in the result file.

***r/system_mysql_db.result	Sun Mar 11 22:51:31 2007
--- r/system_mysql_db.reject	Sun Mar 11 23:29:06 2007
***************
*** 251,255 ****
--- 251,257 ----
    `server_id` int(11) DEFAULT NULL,
    `sql_text` mediumtext NOT NULL
  ) ENGINE=CSV DEFAULT CHARSET=utf8 COMMENT='Slow log'
+ Warnings:
+ Error	1194	Table 'slow_log' is marked as crashed and should be repaired
  show tables;
  Tables_in_test
-------------------------------------------------------


And I assume that is not fixed by the patch?

But, lately we have been seeing that system_mysql_db fails with timeout in PushBuild
and I guess that is what the patch intend to fix? It's strange though that we get these
timeout failures when running with a release compiled mysqld and thus no code from dbug.c
should be involved. We don't pass --debug to mysql-test-run.pl either so how would we need
extend the timeout?

I think there must be some other problem.

Best regards
Magnus


mån 2007-10-15 klockan 19:56 +0300 skrev kgeorge@stripped:
> Below is the list of changes that have just been committed into a local
> 5.1 repository of kgeorge. When kgeorge does a push these changes will
> be propagated to the main repository and, within 24 hours after the
> push, to the public repository.
> For information on how to access the public repository
> see http://dev.mysql.com/doc/mysql/en/installing-source-tree.html
> 
> ChangeSet@stripped, 2007-10-15 19:56:24+03:00, gkodinov@stripped +2 -0
>   Bug #27099: system_mysql fail in pushbuild windows
>   
>   On Windows the debug log was doing freopen () instead of
>   fflush() and that was slowing the logging down that much 
>   that some tests timed out.
>   Fixed by replacing the freopen() with an syncing-to-disk
>   flag to fopen() and fflush().
>   Also increased the timeout of the tests running with --debug
>   on windows : seems to slow down as much as valgrind on linux.
> 
>   dbug/dbug.c@stripped, 2007-10-15 19:56:22+03:00, gkodinov@stripped +10 -18
>     Bug #27099: remove the freopen() for the 
>     windows log : too slow. Replace it with the "c" option
>     for fopen().
> 
>   mysql-test/mysql-test-run.pl@stripped, 2007-10-15 19:56:22+03:00, gkodinov@stripped
> +2 -0
>     Bug #27099: on windows debug log seems to slow the tests
>     down about as much as valgrind does on linux.
> 
> diff -Nrup a/dbug/dbug.c b/dbug/dbug.c
> --- a/dbug/dbug.c	2007-08-02 07:49:24 +03:00
> +++ b/dbug/dbug.c	2007-10-15 19:56:22 +03:00
> @@ -1826,7 +1826,13 @@ static void DBUGOpenFile(CODE_STATE *cs,
>        else
>        {
>          newfile= !EXISTS(name);
> -        if (!(fp= fopen(name, append ? "a+" : "w")))
> +        if (!(fp= fopen(name,
> +#if defined(MSDOS) || defined(__WIN__)
> +		append ? "a+c" : "wc"
> +#else
> +                append ? "a+" : "w"
> +#endif
> +		)))
>          {
>            (void) fprintf(stderr, ERR_OPEN, cs->process, name);
>            perror("");
> @@ -2231,23 +2237,9 @@ static void dbug_flush(CODE_STATE *cs)
>    if (cs->stack->flags & FLUSH_ON_WRITE)
>  #endif
>    {
> -#if defined(MSDOS) || defined(__WIN__)
> -    if (cs->stack->out_file != stdout && cs->stack->out_file !=
> stderr)
> -    {
> -      if (!(freopen(cs->stack->name,"a",cs->stack->out_file)))
> -      {
> -        (void) fprintf(stderr, ERR_OPEN, cs->process, cs->stack->name);
> -        fflush(stderr);
> -        cs->stack->out_file= stderr;
> -      }
> -    }
> -    else
> -#endif
> -    {
> -      (void) fflush(cs->stack->out_file);
> -      if (cs->stack->delay)
> -        (void) Delay(cs->stack->delay);
> -    }
> +    (void) fflush(cs->stack->out_file);
> +    if (cs->stack->delay)
> +      (void) Delay(cs->stack->delay);
>    }
>    if (!cs->locked)
>      pthread_mutex_unlock(&THR_LOCK_dbug);
> diff -Nrup a/mysql-test/mysql-test-run.pl b/mysql-test/mysql-test-run.pl
> --- a/mysql-test/mysql-test-run.pl	2007-10-04 13:16:51 +03:00
> +++ b/mysql-test/mysql-test-run.pl	2007-10-15 19:56:22 +03:00
> @@ -1019,12 +1019,14 @@ sub command_line_setup () {
>    {
>      $opt_testcase_timeout= $default_testcase_timeout;
>      $opt_testcase_timeout*= 10 if $opt_valgrind;
> +    $opt_testcase_timeout*= 10 if ($opt_debug and $glob_win32);
>    }
>  
>    if ( ! $opt_suite_timeout )
>    {
>      $opt_suite_timeout= $default_suite_timeout;
>      $opt_suite_timeout*= 6 if $opt_valgrind;
> +    $opt_suite_timeout*= 6 if ($opt_debug and $glob_win32);
>    }
>  
>    if ( ! $opt_user )
> 
-- 
Magnus Svensson
Senior Software Engineer
msvensson@stripped
+46709164491

Thread
bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099kgeorge15 Oct
  • Re: bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099Magnus Svensson16 Oct
    • Re: bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099Georgi Kodinov16 Oct
      • Re: bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099Magnus Svensson16 Oct
        • Re: bk commit into 5.1 tree (gkodinov:1.2584) BUG#27099Georgi Kodinov16 Oct