[mythtv-users] MythSocket errors, backend unresponsive

[Kevin Ross]

On 08/06/2010 09:08 AM, Bruce Taber wrote:
> On 08/06/2010 11:09 AM, Bob Cottingham wrote:
>    
>> On Thu, Aug 5, 2010 at 6:57 PM, Khanh Tran<khanh at khanh.net>  wrote:
>>      
>>> somewhere around r25084 (I'm running from -fixes branch), I started seeing
>>> the following errors in my master backend:
>>>
>>> 2010-08-05 19:33:37.397 MainServer, Warning: Unknown socket closing
>>> MythSocket(0xffffffffa9634c38)
>>> 2010-08-05 19:33:37.401 MythSocket(ffffffffa9634c38:-1): writeStringList:
>>> Error, socket went unconnected.
>>>                          We wrote 0 of 10 bytes with 1 errors
>>>
>>> The backend process doesn't segfault, but nothing can connect or use it
>>> either.  Sometimes, I get the following errors in the frontend logs:
>>>
>>> 2010-08-05 19:00:46.554 MythSocket(ffffffffb3a6bd10:35): readStringList:
>>> Error, timed out after 30000 ms.
>>> 2010-08-05 19:00:46.554 Connection to backend server lost
>>> 2010-08-05 19:00:46.554 MythContext: Connecting to backend server:
>>> 192.168.1.3:6543 (try 1 of 1)
>>> 2010-08-05 19:00:53.554 MythSocket(ffffffffa4b00c98:35): readStringList:
>>> Error, timed out after 7000 ms.
>>> 2010-08-05 19:00:53.554 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.
>>> 2010-08-05 19:00:53.554 Reconnection to backend server failed
>>>
>>> Anyone else experiencing this?  Have a workaround or fix?
>>>        
>> Are you running a slave backend?  I've been having issues like this
>> since .22-fixes whenever I run my slave backend.  Without the slave
>> backend my master is very stable.  I've seen several people mention
>> issues like this, but never a solution.  I have ticket #7608 open for
>> my issue, but haven't had a single comment.
>>
>> Bob
>> _______________________________________________
>> mythtv-users mailing list
>> mythtv-users at mythtv.org
>> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
>>
>>      
> I've seen this happen running trunk with two slaves. After powering off
> one of the slaves (exit and shutdown) the master no longer responds to
> frontends or status. It will continue recording in-progress recordings.
> If the mythbackend process is killed prior to powering off the slave,
> the master continues to run normally. I have not done any further
> testing to see if something else is influencing the behavior.
>
> Bruce
>    

I get that problem occasionally, too.  Every couple of weeks or so 
(sometimes more often), the slave and master backends stop talking to 
each other.  I need to restart both backends to get them happy again.  
I'm considering consolidating my backends just to avoid this problem.