[mythtv-users] FE Stop/Restart for Smooth Playback

[Stephen P. Villano]

On 3/28/15 12:29 PM, jim wrote:
>
>
> On 03/28/2015 11:51 AM, Mike Perkins wrote:
>> On 28/03/15 14:41, jim wrote:
>>>
>>>
>>> On 03/28/2015 08:48 AM, Mark Perkins wrote:
>>>>
>>>>> On 28 Mar 2015, at 11:03 pm, "jim" <jim_32766 at hotmail.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>>> On 03/27/2015 10:21 AM, jim wrote:
>>>>>> Occasionally, usually after the system has been running 24/7 for
>>>>>> a few days,
>>>>>> playback of recorded television shows is no longer smooth, with
>>>>>> visible
>>>>>> stutter where there should be smooth panning motion. Stopping and
>>>>>> restarting
>>>>>> the FE returns the playback to smooth operation. Has anyone else
>>>>>> experienced
>>>>>> this scenario?
>>>>>> I have experimented with sync, cache drop, and swap stop/restart
>>>>>> to achieve
>>>>>> smooth playback without the FE stop/restart, but only the FE
>>>>>> stop/restart
>>>>>> seems to work. Any ideas or suggestions would be greatly
>>>>>> appreciated. Thanks.
>>>>>>
>>>>>> System Info:
>>>>>> 4 GB Ram
>>>>>> i3 4330 CPU
>>>>>> MythTV Version : v0.27.4-41-gf1115fc
>>>>>> MythTV Branch : fixes/0.27
>>>>>> Network Protocol : 77
>>>>>> Library API : 0.27.20141016-1
>>>>>> QT Version : 4.8.6
>>>>> Based on receiving no answers it may be that nobody else
>>>>> experiences this
>>>>> problem. Please confirm if you do not.
>>>>> My system is very modern, and has plenty of RAM. I wonder what
>>>>> causes this
>>>>> issue?
>>>>>
>>>>> _______________________________________________
>>>> I do not have this problem, my FE can be on for weeks or months at
>>>> a time. But
>>>> I have nvidia graphics, where I assume you use VAAPI.
>>>>
>>>> I would recommend starting by comparing the log file for a
>>>> recording that
>>>> plays fine with that of one that does not to see what the
>>>> differences are.
>>>>
>>> If the difference were vaapi wouldn't the problem be present all the
>>> time? I
>>> wonder if something else running on my system contributes, or if the
>>> OS (Mint
>>> 17.1) is involved. What OS are you running?
>>> I will also check the log file for a recording playback pre/post FE
>>> start/stop.
>>>
>> Any problem like this doesn't automatically have to be continuous.
>> For example, different programs may have different bit rates/frame
>> rates/screen formats. These can impose differing demands on the video
>> driver and hardware.
>>
>> The same also goes for audio, incidentally. If the audio encoding
>> changes mid transmission - say for a commercial break - then that
>> might cause buffering problems which could affect the vodeo.
>>
>> As Mark wrote above, your best bet is to look at the logs, a segment
>> which works and a segment which is bad.
>>
> I use recorded Jeopardy! episodes for testing as the panning of
> categories in Jeopardy! makes for a great test of smooth motion in
> playback. When I test I use the same exact recording, pre and post FE
> stop/start. The FE stop/start completely fixes the problem for some
> inexact period of time. I've not done a good log comparison between a
> smooth playback and a jerky one, and will do so as soon as the
> problems appears again.
>
> I've never noticed a situation wherein the smoothness is better/worse
> before/after a commercial break. I've experimented with SD and HD and
> the full range of standard frame rates. None of that seems to stress
> the system. If the FE has been recently stopped/started all
> combinations are good. The problem behaves more like a buffer/cache
> issue that builds over time and eventually causes a problem. Why FE
> stop/start helps I do not know.
>
> I really appreciate the time people contribute on this mail list
> answering questions. It really helps. Thanks.
>
> _______________________________________________
Try running top during the post restart of the FE and during the poorly
performing ones. That also might give you a clue.
I've had to do that many times, a couple of times, process hogs turned
up. Tuning or stopping unnecessary ones fixed the issue I was having
(non-mythtv system, but the troubleshooting remains the same).