[mythtv-users] mythcommflag taking all the cpu 0.19 svn revision 9806

[Osma Ahvenlampi]

On ke, 2006-05-10 at 20:07 -0400, Khanh Tran wrote:
> Thanks for the help.  While running xine and one mythcommflag, I think
> there's a problem with too many things waiting for CPU time.  The
> higher numbers in the wait column came around when xine paused and I

You have pretty high constant CPU use there, but you also have some idle
CPU time, so for the most part you don't have processes waiting for
that. In fact, the only rows where your CPU idle was zero were those
where you had I/O wait. It's not the CPU that's your problem - it's disk
I/O. The block I/O throughput is at REALLY low levels, under a megabyte
per second, and you should not have any I/O wait time under that load.

Are you sure you have DMA enabled on the hard disk? hdparm -d /dev/hda
to find out. Or maybe your filesystem is fragmented, or badly tuned for
your use, or something. In any event, look at disk as the bottleneck,
not the CPU.

> believe mythcommflag was writing to the database.  There was even the
> end of on recording on mythbackend and the beginning of another.
> 
> Is 0.19 just more taxing on the cpu than 0.18 was?

Anecdotal evidence says sometimes it is, sometimes it isn't. Depends on
which part you look at.

> procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
>  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
>  5  0 119676  22360      0 150884    0    0   756   115  822  1529 74  7  7 12
>  5  0 119676   7800      0 165328    0    0   774     8  800  1470 71 17  0 12
>  1  0 119676   7320      0 166024    0    0   193   663  738  1145 81  3 15  1

-- 
Osma Ahvenlampi   <oa at iki.fi>    http://www.fishpool.org