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

[Warpme]

On 8/5/11 3:16 PM, Jarod Wilson wrote:
> On Aug 5, 2011, at 12:15 AM, Michael T. Dean wrote:
>
>> On 08/04/2011 04:43 PM, Brian J. Murrell wrote:
>>> On 11-08-04 04:39 PM, Michael T. Dean wrote:
>>>> For those of you on
>>>> 0.24-fixes, I will try to create a backport patch and post it here for
>>>> testing later today.
>>> You post it and I will build and run it.
>> I posted a lightly-tested backport on
>> http://code.mythtv.org/trac/ticket/9704 .  You can download it at:
>>
>> http://code.mythtv.org/trac/raw-attachment/ticket/9704/mythtv-0.24-backport_reconnect_fixes.patch
>>
>> It should be applied in the top-level of the mythtv checkout (the
>> directory that contains mythtv and mythplugins directories--since it
>> changes both mythtv and mythplugins):
>>
>> patch -Np1 -i ~/mythtv-0.24-backport_reconnect_fixes.patch
>>
>> Note that because of changes to mythplugins, you will need to recompile
>> plugins.  Also, the patch includes a binary/API version change--so it
>> will enforce the requirement to recompile plugins.  (There are no
>> changes to DB schema version or protocol version, though.)
>>
>> Thanks to all those who test it.
> I can still immediately reproduce the same deadlock I've been seeing 95% of
> the time when I start mythfrontend of late, even with the back-ported patch.
>
> 2011-08-05 09:13:49.857 MythCoreContext: Connecting to backend server: xxx.xxx.xxx.xxx:6543 (try 1 of 1)
> 2011-08-05 09:13:56.861 MythSocket(21f1ea0:50): readStringList: Error, timed out after 7000 ms.
> 2011-08-05 09:13:56.861 Protocol version check failure.
>                          The response to MYTH_PROTO_VERSION was empty.
>                          This happens when the backend is too busy to respond,
>                          or has deadlocked in due to bugs or hardware failure.
> 2011-08-05 09:13:56.861 Unable to determine master backend time zone settings.  If those settings differ from local settings, some functionality will fail.
>
Hi,

Probably this not so helpful - but it think it is worth to mention.
Being on 0.24 I had regularly this type of error.
I believe however only part of them was related to DB.
Other part I believe was related to mythproto server implementation.
At some point in time mythproto server was totally rewritten in master 
and since that time I not experienced single deadlock with
"MYTH_PROTO_VERSION was empty" - beside those which were related to DB.
May You look on stack trace during such deadlock and check is it related 
to DB access hang ?
Call stack usually looks like in my OP.
It think will be helpful as I believe devs prefer to address various 
deadlocks in structured way.
This thread is devoted to DB area and recent Daniel's commits are 
targeted only to this area.
Worst thing may happen here when potentially misleading feedback lead to 
changes in proper code....
Thx

-------------- next part --------------
A non-text attachment was scrubbed...
Name: warpme.vcf
Type: text/x-vcard
Size: 83 bytes
Desc: not available
Url : http://www.mythtv.org/pipermail/mythtv-users/attachments/20110805/d24dfad3/attachment.vcf