[mythtv-users] `Mythbackend leaks memory` continued

Tom Lichti tom at redpepperracing.com
Wed Jul 23 16:00:34 UTC 2008


Udo van den Heuvel wrote:
> Nick Morrott wrote:
>   
>>> Over here the system is running quasi all the time (just three channels
>>> from one multiplex) so stuff never stops recording.
>>> That could explain the figures on my box?
>>>       
>> Maybe. Even when my 2.6.16 backend is recording 3 programmes at once
>> (different multiplexes) memory usage does not continue to rise anymore
>> - so it would seem with my current system configuration (no multirec
>> yet) it is the number of concurrent recordings that influences backend
>> memory usage, not how many recordings are made during the day.
>>     
>
> Can anybody confirm Nick's theory?
>   
I can't confirm it, but I am seeing a leak as well. No EIT, no multi-rec 
here. Over the last 8 days, I have had 47 recordings made, and the 
mythbackend usage went from:
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
21401 root       7 -15  234m  30m 8920 S    0  3.1   0:10.93 mythbackend

to:

21401 root       7 -15  282m  62m  10m S    0  6.3  54:18.08 mythbackend

I am taking those stats hourly. The high VIRT was 294m, and high RES was 
74m, so it is cleaning up some stuff. I will try valgrind at some point 
and see if I can get anything useful.

> In combination with my way of recording that could explain the figures I
> see here.
>   
Perhaps. Or it's the same problem just magnified by your environment.
> On the other hand Andrew's numbers are from a different usage pattern,
> so maybe a different leak?
>   

See previous statement.


Tom



More information about the mythtv-users mailing list