List:Commits« Previous MessageNext Message »
From:Christopher Powers Date:June 15 2010 8:08pm
Subject:Re: bzr commit into mysql-next-mr-bugfixing branch (marc.alff:3245)
Bug#54467
View as plain text  
Ok to push.

On 6/15/10 2:06 PM, Marc Alff wrote:
> #At file:///Users/malff/BZR_TREE/mysql-next-mr-bugfixing-54467/ based on
> revid:alik@stripped
>
>   3245 Marc Alff	2010-06-15
>        Bug#54467 performance schema complains of wrong structure in bootstrap mode
>
>        Prior to this fix, the installation script:
>        - did print [Note] messages in the server log,
>          related to the performance schema.
>
>        - did print [ERROR] messages in the server log,
>          complaining about performance schema tables.
>          These errors are not justified during bootstrap.
>
>        With this fix:
>        - notes are no longer printed,
>          this reduces the server verbosity.
>
>        - integrity checks for the performance schema tables
>          are not performed when tables are not expected to be there,
>          to avoid un justified [ERROR] messages.
>
>        Tested the mysql_install_db script manually:
>
>        Installing MySQL system tables...
>        OK
>        Filling help tables...
>        OK
>
>      modified:
>        sql/mysqld.cc
> === modified file 'sql/mysqld.cc'
> --- a/sql/mysqld.cc	2010-06-04 20:20:13 +0000
> +++ b/sql/mysqld.cc	2010-06-15 19:06:04 +0000
> @@ -782,7 +782,15 @@ void Buffered_log::print()
>       sql_print_warning("Buffered warning: %s\n", m_message.c_ptr_safe());
>       break;
>     case INFORMATION_LEVEL:
> -    sql_print_information("Buffered information: %s\n", m_message.c_ptr_safe());
> +    /*
> +      Messages printed as "information" still end up in the mysqld *error* log,
> +      but with a [Note] tag instead of an [ERROR] tag.
> +      While this is probably fine for a human reading the log,
> +      it is upsetting existing automated scripts used to parse logs,
> +      because such scripts are likely to not already handle [Note] properly.
> +      INFORMATION_LEVEL messages are simply silenced, on purpose,
> +      to avoid un needed verbosity.
> +    */
>       break;
>     }
>   }
> @@ -4540,16 +4548,6 @@ int mysqld_main(int argc, char **argv)
>           buffered_logs.buffer(WARNING_LEVEL,
>                                "Performance schema disabled (reason: init
> failed).");
>         }
> -      else
> -      {
> -        buffered_logs.buffer(INFORMATION_LEVEL,
> -                             "Performance schema enabled.");
> -      }
> -    }
> -    else
> -    {
> -      buffered_logs.buffer(INFORMATION_LEVEL,
> -                           "Performance schema disabled (reason: start
> parameters).");
>       }
>     }
>   #else
> @@ -4817,7 +4815,14 @@ int mysqld_main(int argc, char **argv)
>
>   #ifdef WITH_PERFSCHEMA_STORAGE_ENGINE
>     initialize_performance_schema_acl(opt_bootstrap);
> -  check_performance_schema();
> +  /*
> +    Do not check the structure of the performance schema tables
> +    during bootstrap:
> +    - the tables are not supposed to exist yet, bootstrap will create them
> +    - a check would print spurious error messages
> +  */
> +  if (! opt_bootstrap)
> +    check_performance_schema();
>   #endif
>
>     initialize_information_schema_acl();
>
>
>
>
>
Thread
bzr commit into mysql-next-mr-bugfixing branch (marc.alff:3245) Bug#54467Marc Alff15 Jun
  • Re: bzr commit into mysql-next-mr-bugfixing branch (marc.alff:3245)Bug#54467Christopher Powers15 Jun