[mythtv-users] Crippling load spike on FE/BE - cause unknown?

Anthony Messina amessina at messinet.com
Fri Jan 2 10:57:13 UTC 2009


On Monday 08 December 2008 11:21:34 Anthony Messina wrote:
> I've been having difficulty with my freshly build F10 x86_64 (using a
> previous mythconverg db that I was using in F9, F8, F7 ...) MythTV FE/BE
> box and CPU load issues for a few months.  I've so far, been unsuccessful
> at getting the cause nailed down on my own. The issue was happening with F9
> as well as F10.
>
> I am using mythtv-0.21-fixes from livna (though the same happened with
> atrpms) and I do use the nvidia-rpms from livna.
>
> The basic problem is that for some unknown reason, the load spikes WAY up,
> the display freezes, refuses to accept lirc/kb/mouse input, but I am able
> to SSH in and do whatever I want without any delay--basically, it doesn't
> act sluggish via ssh.
>
> Also, during the "lockups", no errors are reported to /var/log/messages,
> dmesg, or the myth logs.  No kerneloops messages.  It does not happen
> during playback or any interactive usage--only when no one is using it
> interactively. The backend continues to record, and function properly
> during the lockups.  I can access mythweb from other boxes without issue.
>  I can access and stream shows (even HD) to another FE without issue.
>
> I've tried various things: disabling as many extra services as possible,
> switching irqpoll on/off, XvMC on/off, switching the recording filesystems
> from jfs back to ext3 with slow delete, changing the commflag cpu usage
> from high to med to low (low actually seemed to make it worse), not using
> mythwelcome/using mythwelcome, selinux on/off. i do not transcode and i do
> not have "realtime priority enabled" as i run the FE under a user accoutn
> just for myth.
>
> The load does not spike in relationship to a mythfilldatabase run or a
> commflag, or the start of the recording. it "seems" totally random. In
> order to restore the system to a workable state, I must restart it, then
> it's fine for a while, then the load shoots up again.
>
> In short, I would appreciate any kind of pointers on where to go--the WAF
> is dropping very quickly.

It appears that using --noupnp has solved the problem, at least for about a 
week.  Not sure why though.  I had two directories set up over nfsv4 to hold 
videos (not recordings).

Thanks to those who tried to help.  -A

-- 
Anthony - http://messinet.com - http://messinet.com/~amessina/gallery
8F89 5E72 8DF0 BCF0 10BE 9967 92DC 35DC B001 4A4E

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url : http://mythtv.org/pipermail/mythtv-users/attachments/20090102/2369df08/attachment.pgp 


More information about the mythtv-users mailing list