[mythtv-users] `Mythbackend leaks memory` continued

Nick Morrott knowledgejunkie at gmail.com
Fri Jul 25 12:09:10 UTC 2008


On 25/07/2008, Ray Lischner <linux at tempest-sw.com> wrote:
> On Wednesday 23 July 2008 06:40 pm, Nick Morrott wrote:
>
>  > To make my previous comment a little clearer, each recording adds a
>  > chunk of RSS to the mythbackend process, which is released once the
>  > recording finishes.
>
>
> RSS is not significant. RSS (Resident Set Size) says how much physical
>  RAM the operating system has given the process. The number will go up
>  or down as other processes make claims for their own memory.

Ray,

Thanks for the detailed reply. With my particular configuration it
seems that disabling active EIT scanning (passive scanning still
occurs during recording) was enough to stop the rising memory use -
nothing else on the box (which also runs other servers) has changed so
at least in my case I feel somewhat confident in attributing the
cause. With active EIT enabled, VSZ was some 50% higher (and still
increasing) than with it disabled after the same 8 day period after
restarting mythbackend.

I will endeavour to try both valgrind and enabling active DVB on a
per-card basis to see whether the increasing memory growth
demonstrated in my graph occurs uniformly whenever active EIT is
enabled (for any number of cards), or whether it increases in
proportion to the number of cards having active EIT enabled.

With Le Tour finishing this weekend, I should be able to look into it
early next week and hope to be able to provide more definitive
evidence regarding the memory growth.

-- 
Nick Morrott

MythTV Official wiki:
http://mythtv.org/wiki/
MythTV users list archive:
http://www.gossamer-threads.com/lists/mythtv/users

"An investment in knowledge always pays the best interest." - Benjamin Franklin


More information about the mythtv-users mailing list