[mythtv-users] Random Stuttering

[George Mari]

Bryce wrote:
> Hi All,
> 
> I have an issue with my myth system that i'm struggling to get working.
> 
> I have a backend system thats an AMD Semperon 3000+ with 512MB Ram
> 10Gig boot disk and 2x200GB disks for media.
> This has a Dvico DVB card for free to air TV (in australia) and a PVR-500
> for foxtel.
> 

Are your 2x200GB disks IDE or SATA?

> I then have a frontend system thats a Pentium D 2.8Ghz, 1Gb Ram
> 
> The problem i get is that I get random stuttering of video recording.  It
> might work fine for 30 mins then 2-3 seconds of stuttering or I might get
> stuttering ever few minutes.  It seems random.
> 
> I can't see anything happening out of the ordinary on either system when
> this occurs.  the only thing is that if i have a terminal window open on my
> backend system and ping the frontend, during the stutering the ping times
> jump up to over 1000ms for that few seconds - but the terminal windows on
> the box don't freeze or anything like that.
> 
> 64 bytes from 192.168.16.11: icmp_seq=249 ttl=64 time=0.326 ms
> 64 bytes from 192.168.16.11: icmp_seq=250 ttl=64 time=2156 ms
> 64 bytes from 192.168.16.11: icmp_seq=251 ttl=64 time=1156 ms
> 64 bytes from 192.168.16.11: icmp_seq=252 ttl=64 time=156 ms
> 64 bytes from 192.168.16.11: icmp_seq=253 ttl=64 time=0.132 ms
> 
> But pinging other boxes, or pinging the backend from a 3rd box does not do
> this.
> 
> I've also tried to play the videos via samba on my windows machine and it
> seems that its the actual videos that have the stuttering in them, not the
> playback, so that makes me think its probably a backend problem, not a
> frontend problem.
> 
> The stutering happens on both high definition and foxtel.
> 
> The backend does not seem to be working too hard:
> 
> 
> top - 13:49:51 up 14:49,  1 user,  load average: 0.20, 0.10, 0.06
> Tasks:  58 total,   2 running,  56 sleeping,   0 stopped,   0 zombie
> Cpu(s):  0.3%us,  0.7%sy,  0.0%ni, 96.7%id,  0.7%wa,  1.0%hi,  0.7%si,
> 0.0%st
> Mem:    451472k total,   446396k used,     5076k free,     1356k buffers
> Swap:   500464k total,      116k used,   500348k free,   356172k cached
> 
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>  4569 mythtv    18   0  251m  20m  13m S  0.7  4.7   1:46.77 mythbackend
>  4360 mysql     15   0  140m  29m 4744 S  0.0  6.8   0:04.55 mysqld
>    71 root      10  -5     0    0    0 S  0.0  0.0   0:01.06 kswapd0
>   784 root      21  -4  1908  704  376 S  0.0  0.2   0:00.84 udevd
>  6568 root      23   0     0    0    0 S  0.0  0.0   0:00.55 nfsd
>  6564 root      15   0     0    0    0 S  0.0  0.0   0:00.51 nfsd
>  6565 root      15   0     0    0    0 S  0.0  0.0   0:00.44 nfsd
> 
> I dont' see major conficts for interuppts
> 
> wolverine ~ # cat /proc/interrupts
>            CPU0
>   0:   13429809    XT-PIC-XT        timer
>   1:          8    XT-PIC-XT        i8042
>   2:          0    XT-PIC-XT        cascade
>   5:     519423    XT-PIC-XT        ivtv0
>   7:      89130    XT-PIC-XT        cx88[0], cx88[0]
>   8:          2    XT-PIC-XT        rtc
>  10:          0    XT-PIC-XT        ivtv1
>  11:    5155231    XT-PIC-XT        sata_via, eth0
>  12:        130    XT-PIC-XT        i8042
>  14:      34189    XT-PIC-XT        ide0
>  15:      65054    XT-PIC-XT        ide1
> NMI:          0
> LOC:   13430083
> ERR:          0
> Any advice or suggestions would be greatly appreciated.
> 

If your media drives are SATA, the fact that your SATA controller shares 
an interrupt with your ethernet device might explain your high ping 
times during the episodes of stuttering, but may not explain the cause 
of the stuttering in the first place.  Any other network-intensive 
activity happening on your backend during recording?  I see it's running 
  the NFS daemon - any other boxes writing or reading a lot of data to 
those drives?  If so, you may want to work on separating those two 
devices so they do not share an interrupt.

If your media devices are IDE, are you sure DMA is turned on?  (See the 
hdparm utility)

What level of logging are doing with mythbackend?  I have an 
underpowered backend that could not cleanly serve up one video stream to 
a remote frontend client with mythbackend running verbose -all logging. 
  If I turned off all logging, it was fine.  I've since upgraded the 
disk I/O capabilities of the backend.