Re: [Maria-developers] deadlock, rotate and prepared xids

2013-03-05 Thread Kristian Nielsen
Kristian Nielsen writes: >> http://bugs.mysql.com/bug.php?id=>68569 > > This is the classic semi-sync deadlock that you explained so many times > already. >> http://bugs.mysql.com/bug.php?id=68251 > > This second one is different from the first one. One thread is trying to I mixed up the URLs.

Re: [Maria-developers] deadlock, rotate and prepared xids

2013-03-05 Thread MARK CALLAGHAN
Thanks for the analysis. Filed http://bugs.mysql.com/bug.php?id=68575 for the frequent rotate problem. On Tue, Mar 5, 2013 at 2:02 AM, Kristian Nielsen wrote: > MARK CALLAGHAN writes: > > > http://bugs.mysql.com/bug.php?id=68569 > > This is the classic semi-sync deadlock that you explained so m

Re: [Maria-developers] deadlock, rotate and prepared xids

2013-03-05 Thread Kristian Nielsen
MARK CALLAGHAN writes: > http://bugs.mysql.com/bug.php?id=68569 This is the classic semi-sync deadlock that you explained so many times already. The 5.6 group commit implements a similar technique to MariaDB, where a single thread does the fast part of commit for all transactions in the group.

[Maria-developers] deadlock, rotate and prepared xids

2013-03-04 Thread MARK CALLAGHAN
Kristian, Did you do anything special to avoid deadlock with ::new_file_impl and prepared XIDs? My memory is vague but I thought you handled it. I added code to the FB patch group commit implementation to handle it. Both Yoshi and I hit deadlocks with that code in official MySQL 5.6: http://bugs.m