List:Commits« Previous MessageNext Message »
From:Ingo Strüwing Date:May 6 2009 9:18am
Subject:Re: bzr commit into mysql-6.0-backup branch (charles.bell:2803)
Bug#39924
View as plain text  
Hi Chuck,

thanks for applying the proposed changes.

Status: approved

Thoughts:
[1] That's ok in itself. However, I think it might be more similar to
the treatment of other variables, if we copy its value at the top of the
test case, not at the top of each test. OTOH, your approach is better
for those, who insert temporary changes to the variable for debugging
purposes. I'm undecided, which way to prefer. Perhaps you better leave
it as you have it now.

Details:
Chuck Bell, 05.05.2009 23:32:

...
>  2803 Chuck Bell	2009-05-05
>       BUG#39924 : Not possible to use backup logs after BACKUP has been interrupted
>       
>       When a backup or restore was cancelled or interrupted and the thread
>       was killed, the backup logs were left in an undefined state. This was
>       because they were opened using a method that detects a killed thread
>       and fails if the thread is in a killed state.
>       
>       This patch includes an override in the table open method that allows
>       the backup logs to be written if the thread is in a killed state.
>       
>       Before the patch the logs were incomplete with the error condition
>       missing from the progress log and the history log entry either 
>       incomplete or not written at all. 
>       
>       This patch ensures the log entries will be written in the event the 
>       thread is killed (backup or restore is cancelled in some unusual
>       manner).
...
> === modified file 'mysql-test/suite/backup/t/backup_logs.test'
...
> +--echo #
> +--echo # Test to see if logs successfully written after command is 
> +--echo # killed.
> +--echo #
> +let $debug= `SELECT @@debug`;
> +SET SESSION debug="+d,kill_backup";


[1] That's ok in itself. However, I think it might be more similar to
the treatment of other variables, if we copy its value at the top of the
test case, not at the top of each test. OTOH, your approach is better
for those, who insert temporary changes to the variable for debugging
purposes. I'm undecided, which way to prefer. Perhaps you better leave
it as you have it now.

...

Regards
Ingo
-- 
Ingo Strüwing, Database Group
Sun Microsystems GmbH, Sonnenallee 1, D-85551 Kirchheim-Heimstetten
Geschäftsführer: Thomas Schröder,   Wolfgang Engels,   Wolf Frenkel
Vorsitzender des Aufsichtsrates: Martin Häring   HRB München 161028
Thread
bzr commit into mysql-6.0-backup branch (charles.bell:2803) Bug#39924Chuck Bell5 May 2009
  • Re: bzr commit into mysql-6.0-backup branch (charles.bell:2803)Bug#39924Ingo Strüwing6 May 2009