[mythtv-users] increasing load

Udo van den Heuvel udovdh at xs4all.nl
Tue May 13 19:30:42 UTC 2008


Jan Ceuleers wrote:
>> Udo van den Heuvel wrote:
>>  Any ideas why the load could be rising?
>>  Or how I could debug this?
> 
> On the subject of rising memory consumption, if you are using a tmpfs 
> mount on /tmp, that might explain part of it.

That is not the case:
# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda5             950M  350M  551M  39% /
/dev/sda1              99M   38M   57M  40% /boot
tmpfs                 475M     0  475M   0% /dev/shm
/dev/sda6             487M   44M  428M  10% /home
/dev/sda8             4.3G  3.8G  386M  91% /usr
/dev/sda7             872M  452M  385M  55% /var
/dev/sda10            459G  455G  3.7G 100% /Myth

Restarting mythbackend helped and could also explain the memory leak
hint that was posted.

> On the increasing load: can I suggest that you take regular snapshots of 
> files that contain statistical info (e.g. /proc/interrupts, 
> /proc/meminfo) and see if that turns up any useful leads.

powertop says mythbackend causes the majority of interrupts from C2
state. (about 55% for do_nanosleep (hrtimer_wakeup) )
Wakeups-from-idle per second : 527.3
Is that normal?

> Another thing: if you are using CPU frequency scaling (as I am) then the 
> CPU load can vary in strange ways.

Don't use scaling as I only have 1.2 Ghz in that tiny system. Also in
the past the scaling wasn't too stable as it is in my other systems.

> Clasping at straws here, but hope it helps.

Thanks for posting your suggestions/ideas.


Kind regards,
Udo


More information about the mythtv-users mailing list