[mythtv-users] Waiting for video buffers

[Joe Henley]

I've had this problem ALOT.  I've also done alot of testing on it.  What 
I've found to reduce/eliminate it on my system (running FC16 with kernel 
at 3.6.11-1, MythTV 0.25.2, MiniMyth 0.25.2-288) includes:
- In general the problem is 95% in playback, but a small portion is 
created during recording and then passed on to display the issue during 
playback.

- In playback the biggest source of the message is using fast forward 
and/or rewind, pause, start/stop play (ie., not really a problem).
The next biggest source is the selection of deinterlacer used.  Even 
with vdpau in use, I find 1X deinterlacers produce fewer problems (than 
2X).  The lesson I learned is that I should not be aggressive with this 
setting.

- In recording, the biggest source of (my .... very specific) problem is 
leaving the PVR-350 card in the PCI slot even though I'm not using it.  
That card uses the ivtv module which is in the kernel.  So even with 
ivtv blacklisted, and the card not used, it causes me alot of these 
video buffer messages.  The typical problem with the ivtv module 
(tearing, etc.) "gets you" even if you're not using it (see 
http://code.mythtv.org/trac/ticket/10732).  [I know this sounds 
implausible, but I have tested it every which way I can, and with the 
card in the slot (even changing slots to change interrupts) I get the 
video buffer messages and with it removed, I don't.]
The next biggest source is a bad signal from my cable provider (of 
course, it's Time Warner).  Any time you see a brief stutter, or garbled 
signal, on the cable feed (I'm not talking about LiveTV via Myth), then 
if you're recording it, there will be lots of video buffer messages.  
Thus, since you seldom watch the cable feed of a program you're 
recording, you tend to assume it comes from your Myth system.

There are a couple of other, in my case lesser, issues:
-  I used two nics in my Myth server, set as multi homed.  Using NFS in 
that environment did cause some video buffer messages, but not alot.  
Even still, I eliminated the second nic, eliminated the multi homing, 
but kept NFS.
- I increased the  ringbuffer from 18,800 kB  to 28,200.   That seems to 
help reduce capture problems (which eventually show up in the FE as 
video buffer messages).

As always, YMMV.  But I hope this gives you some ideas as to where to 
look to sort out the problems you are having.  Good luck and let us know 
how you do.

Joe Henley