[mythtv-users] Playback problem -- random short pauses

[Steven Adeff]

On Tue, May 17, 2011 at 2:43 PM, Kenneth Emerson
<kenneth.emerson at gmail.com> wrote:
> On Tue, May 17, 2011 at 11:49 AM, Steven Adeff <adeffs.mythtv at gmail.com>
> wrote:
>> On Fri, May 6, 2011 at 9:46 AM, Kenneth Emerson
>> <kenneth.emerson at gmail.com> wrote:
>> > Last night I modified the routine RingBuffer.cpp
>> > (method: RingBuffer::CalcReadAheadThresh) to always choose the largest
>> > read
>> > block size.  Whether this was a proper change or not it did seem to
>> > eliminate the pauses I had been seeing.  I watched approximately four
>> > hours
>> > of recordings without any pausing of the video or audio.  I will leave
>> > this
>> > change in for a week to see if it has really solved the symptoms (if not
>> > the
>> > problem). At that point I might open a ticket with the patch.
>> > Here is the diff for anyone else brave enough to give it a try:
>> > diff --git a/mythtv/libs/libmythtv/RingBuffer.cpp
>> > b/mythtv/libs/libmythtv/RingBu
>> > index 8b59616..d4613d6 100644
>> > --- a/mythtv/libs/libmythtv/RingBuffer.cpp
>> > +++ b/mythtv/libs/libmythtv/RingBuffer.cpp
>> > @@ -816,7 +816,10 @@ void RingBuffer::CalcReadAheadThresh(void)
>> >      rbs            = (estbitrate > 5000)  ? KB128 : rbs;
>> >      rbs            = (estbitrate > 9000)  ? KB256 : rbs;
>> >      rbs            = (estbitrate > 18000) ? KB512 : rbs;
>> > -    readblocksize  = max(rbs,readblocksize);
>> > +    // readblocksize  = max(rbs,readblocksize); Ticket #8706:
>> > 8706_debug.diff
>> > +    //  Try to maximize readahead buffer. By KRE 05/04/2011
>> > +    rbs = KB512;
>> > +    readblocksize  = rbs;
>> >      // minumum seconds of buffering before allowing read
>> >      float secs_min = 0.25;
>> > Regards,
>> > Ken E.
>>
>> Ken, any word on how the patch worked out after a week of use?
>>
> Unfortunately, this patch provided no relief.  I have since backed out the
> change. I haven't been diligent in trying to correlate all of the pauses to
> external events, but the ones that I have chased down seem to align with
> other commenters in this thread.  I definitely see pauses when recordings
> start/stop in the backend.  I also see pauses when auto-expire occurs and
> files get deleted as well as when the preview image is being constructed.  I
> do have the WD green drives, but I checked the SMART parameters for head
> parking and that is not the issue (slightly more head parks than power
> cycles and < 100 for three year old drives).
> I still believe that something has changed as this wasn't a problem until
> sometime in .24-fixes (guessing within the last three months).  Someone in
> this thread suggested going back to when .24 was forked off (last db
> update).  Kenni and Jean-Yves both seemed to have tried going backward and
> were not able to locate where the problem surfaced (thinking it was pre .24
> release).
> I am on the verge of upgrading to 0.25 head hoping that most of the core
> changes for the 0.25 release have been made and that head is somewhat stable
> as this is the family's only DVR.  It also might be useful if people who are
> seeing this problem report their configuration to see if there is any
> commonality in hardware and/or software.  I am assuming that not everyone
> is experiencing this problem.  So I will start:
> Hardware:
>     Motherboard:  P5K-E/WIFI-AP LGA 775
>     CPU: Core 2 Quad Q6600 Kentsfield 2.4GHz
>     Memory:  4GB DDR2 1066
>     Disk:  4 x Western Digital Caviar Green WD10EACS 1TB SATA 3.0Gb/s
> (sda/b/c/d)
>               1 x  Western Digital WD5000KS 500GB SATA (sdf)
>     Video:
>     TV: Hauppauge HVR-1800 PCIe
>            Hauppauge PVR-250  PCI
>            Hauppauge PVR-350 PCI
>>
>>      HDHR dual tuner
>
>         Hauppauge HDPVR v 0x12
> Software
>      OS: Ubuntu 10.04.02 LTS (32 bit)
>      Kernel: 2.6.32-31-generic #61-Ubuntu SMP
>      Other packages:
>          exim4 mail server
>          apache2 web server
>          samba v3.4.7
>          smartd v5.38
>          all mythtv scripts run by cron between 3am and 5am except for
> mythfilldatabase and mythcal which are run back-to-back between 5pm and 6pm.
>      MythTV combined FE/BE (no other computers)
>      Drive partition:
>      sda1/sdb1/sdc1/sdd1 -> RAID1 boot ext3
>      sda2/sdb2/sdc2/sdd2 -> RAID1 root ext3
>      sda3/sdb3/sdc3/sdd3 -> RAID1 swap
>      sda4/sdb4/sdc4/sdd4 -> RAID5 LVM xfs (recordings)
>      sdf1 -> mysql database
>     mythfrontend running from xinit and gnome
>     mythbackend running with debug v0.24.1-1-g347cd24-dirty (in the process
> of removing all edits)
>
>>
>> -- Ken E.

bummer. it's occuring on my parents system. they only have two tuners
and two frontends (it's a combined BE/FE and a standalone FE). they
don't have any Western Digital drives though, and the ones they do
have don't have any SMART issues. They have three recording drives and
there isn't a direct correlation between wa% and the hiccups (the
machine basically never has a wa% above 20%).

I've noticed it a few times on my Atom frontends (though they don't
seem to show anything in their logs with -v playback), but my
Athlon/GT220 frontend only sometimes when many shows start/stop
recording on the backend (which is a standalone). my systems can be
seen on my wiki page below in my sig.

I'm also one of those folks who had no issues like this with 0.23 but
had them creep up with 0.24.

-- 
Steve
http://www.mythtv.org/wiki/User:Steveadeff
Before you ask, read the FAQ!
http://www.mythtv.org/wiki/index.php/Frequently_Asked_Questions
then search the Wiki, and this list,
http://www.gossamer-threads.com/lists/mythtv/
Mailinglist etiquette -
http://www.mythtv.org/wiki/index.php/Mailing_List_etiquette