[mythtv] MPEG2->MPEG2 Transcode Errored (0.19 fixes)

Thomas Börkel thomas at boerkel.de
Tue Aug 8 21:32:14 UTC 2006


HI!

Geoffrey Hausheer wrote:

> Quite possible, but extremely odd.  when the mpeg2-transcoder kicks in
> it really has no dependance on the arguments used (unless you have set
> the '--ostream' switch).  In any case, Thomas, it looks like I was
> worng, and sent you off i the wrong direction.  Sorry about that.

No problem.

> Looking at the logs, something weird did happen around position
> 1979582472.  It stopped processing frames, and just started adding

Which log? mythbackend.log or the stand-alone mythtranscode.log?

> them to the queue.  My guess is that the process ran out of memory and
> got killed either by the OOM killer or segfault.  It crashes around
> byte 2239288304 which is 260MB of queued up frames (far more than

But why did it run through when started stand-alone? The result looks 
OK. The original is 2.1 GB and the transcoded (without commercials) is 
963 MB.

And the error when running from mythbackend and the success when running 
stand-alone is both reproducable with the recording.

> should ever happen) (you can tell this is happening by the lack of the
> AUD/VID messages in the log.  here is something for you to try:
> 
> dd if=/work2/video/recordings//3649_20060801220000.mpg
> of=/work2/small.mpg bs=1000000 skip=1970 count=25
> 
> then run mythtranscode on that with full logs.  It probably won't
> crash, but if you see that the AUD/VID messages start at the
> beginning, then stop halfway through, that is good enough.  In which
> case, I'll want that that test-stream.  If in doubt, just send me the
> logs, and I'll tell you whether it is sufficient or not.

http://www.boerkel.de/mythtranscode.log.gz

http://www.boerkel.de/small.mpg

Thanks!

Thomas



More information about the mythtv-dev mailing list