[mythtv-users] backend stops responding 2-3 times a day

[Brian J. Murrell]

On Sun, 2011-01-23 at 14:47 +0000, Alex Tomlins wrote: 
> Hi all,

Hey,

> I'm having an issue where mythbackend will stop responding 2-3 times a 
> day.

Welcome to the club.

> When this happens the process is still running, and logging stuff, 
> but none of the frontends can connect, mythweb stalls, and the mythvideo 
> cronjobs all fail.

Yup.

> Additionally any recordings fail.  Restarting 
> mythbackend always gets things back up and running again.

~nods~

> I'm running a dedicated backend with 2 Hauppauge Nova-T capture cards, 
> running ubuntu 10.04 with the mythbunto autobuilds.

What are Nova-T capture cards?  Do they capture clearqam or are they
analog signal encoders?  Which kernel driver do you use with them?

> The backend log at the time it fails contains:
> 
> 2011-01-23 10:11:02.674 MythSocket(ffffffffa932f890:-1): 
> writeStringList: Error, socket went unconnected.
>                          We wrote 0 of 9 bytes with 1 errors

And there it is.

> When this happens the frontend log contains:
> 
> 2011-01-23 05:59:01.600 MythSocket(ac0dc98:45): readStringList: Error, 
> timed out after 300
> 00 ms.
> 2011-01-23 05:59:01.601 Connection to backend server lost
> 2011-01-23 05:59:01.601 MythCoreContext: Connecting to backend server: 
> 10.10.10.2:6543 (tr
> y 1 of 1)
> 2011-01-23 05:59:08.602 MythSocket(c1be660:45): readStringList: Error, 
> timed out after 7000 ms.
> 2011-01-23 05:59:08.602 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.
> 2011-01-23 05:59:08.602 Reconnection to backend server failed
> 2011-01-23 05:59:08.603 MythCoreContext: Connecting to backend server: 
> 10.10.10.2:6543 (try 1 of 1)
> 2011-01-23 05:59:08.812 New DB connection, total: 1
> 2011-01-23 05:59:08.814 Connected to database 'mythconverg' at host: 
> galadriel.tomlins.org.uk
> 2011-01-23 05:59:15.607 MythSocket(ffffffffa1112ef0:45): readStringList: 
> Error, timed out after 7000 ms.
> 2011-01-23 05:59:15.607 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.
> (last 3 lines repeated many times until backend is restarted)

Yup.

> The last time it failed I attached gdb to it to obtain a backtrace as 
> per instructions 
> http://www.gossamer-threads.com/lists/mythtv/users/455557#455557.  
> backtrace attached.

I have obtained a number of gdb traces as was requested the last time
around on this one and put them into bug 9274.  Unfortunately that bug
has not gotten any real attention at all other than to chastise about
pasting stack traces rather than attaching them.

> Can anyone give some pointers as to where to look further.

http://code.mythtv.org/trac/ticket/9274?  That looks like what you are
seeing.

b.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://mythtv.org/pipermail/mythtv-users/attachments/20110123/af9cca42/attachment-0001.pgp>