[mythtv-users] MySQL related BE deadlocks - collective wisdom needed

[Michael T. Dean]

On 07/27/2011 03:09 PM, Brian J. Murrell wrote:
> On 11-07-27 03:01 PM, Michael T. Dean wrote:
>> A better test would be to apply
>> https://github.com/MythTV/mythtv/commit/dd2d7a14d to your MythTV (or, if
>> you're running master, just pull an update).  If you're on 0.24-fixes,
>> it will be much easier to backport this change than it was to revert the
>> commit I mentioned.  Just make sure you're using Qt 4.6 or higher if
>> you're on 0.24-fixes.
> Is all of this to indicate that this won't land on 0.24-fixes?
>
> I've been following this thread with great interest since I think it's
> the cause of #9704 also, which has been plaguing me for much too long.

Since 0.24-fixes requires Qt 4.5+, we can't use this on it.

OK, actually, we kind of could, but the logs would be littered with

QMYSQLDriver::open: Illegal connect option value 'MYSQL_OPT_RECONNECT=1'

or something like that for anyone using Qt4.5.  In theory, it won't 
cause any problems for them, but until we have some reasonable proof 
that it's actually fixing the issue, we'll  just test the code in 
unstable.  If we do find evidence that it actually works to prevent 
these deadlocks, I'll see about adding it to 0.24-fixes--so at least 
people using Qt 4.6+ will benefit from it.

Input from those of you experiencing the deadlocks would make it much 
easier.  So, if you can, please try building your MythTV with the patch 
and using it.

I've attached a version of the patch for 0.24-fixes.

Thanks,
Mike
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mythtv-0.24-fixes-set_MYSQL_OPT_RECONNECT.patch
Url: http://www.mythtv.org/pipermail/mythtv-users/attachments/20110727/fb413003/attachment.pl