<div><span class="gmail_quote">On 9/20/05, <b class="gmail_sendername">Mark Knecht</b> <<a href="mailto:markknecht@gmail.com">markknecht@gmail.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On 9/20/05, Mercury Morris <<a href="mailto:mercury.morris@gmail.com">mercury.morris@gmail.com</a>> wrote:<br>> On 9/19/05, Michael T. Dean <<a href="mailto:mtdean@thirdcontact.com">mtdean@thirdcontact.com</a>
> wrote:<br>> > Could this be a problem with your system (instead of a problem with<br>> > Myth) that others are seeing where running mythfilldatabase (or wget of<br>> > any kind) while recording causes an IOBOUND situation that prevents
<br>> > information from being written properly and causes an A/V sync offset?<br>> > In other words, does Charlie Rose happen to record around the time your<br>> > mythfilldatabase runs?<br>> ><br>
> ><br>> <a href="http://www.gossamer-threads.com/lists/mythtv/commits/150176">http://www.gossamer-threads.com/lists/mythtv/commits/150176</a><br>><br>> Yes Mike ! In the mythbackend.log, there are IOBOUND entries that
<br>> correspond in time exactly! to the discontinuities in the Charlie Rose<br>> show.<br>><br>> But here's the thing: I had already suspected mythfilldatabase because<br>> it had been running within the Charlie Rose hour. So, I changed the
<br>> window for mythfilldatabase to run outside of that hour - problem still<br>> occurred.<br><br>Did you completely restart the backend mysql processes after doing<br>this? Maybe even reboot to be sure?</blockquote>
<div><br>
Yes. After changing the mythfilldatabase window, I waited until two<br>
days had passed. During those two days, the system did its normal<br>
shutdown-and-wakeup thing. That is, it shutdown the computer and<br>
re-booted at the times when programs were scheduled to be recorded.<br>
<br>
</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">><br>> So, mythfilldatabase is not the cause, BUT the system is recording<br>> Letterman (in HDTV) from an HD3000 card at the same time.
<br>> I guess that the simultaneous recordings are too much for the I/O<br>> system to handle - a bit of a surprise to me.<br><br>Not likely unless you have no DMA on your hard drives or the data is<br>traversing the network. My backend storage is not on my backend
<br>server. I never see problems like this but I don't record HD either.<br>None the less that should be less than 5MB/S which even a non-DMA hard<br>drive might keep up with...</blockquote><div><br>
Yeah, that's what I thought ! How could the I/O system of newly-built<br>
computer be unable to keep up with only two write-to-disk tasks. I've<br>
watched the disk-activity light (not a good measure of load, I know),<br>
and it blinks slowly. It's nothing like when the system is running updatedb<br>
or makewhatis. <br>
</div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">><br>> But now, thanks to you pointing me to the logs, I can set up the<br>
> recording schedules so that only one recording takes place at a time.<br>> (Of course, I feel that I should have looked in the logs without having<br>> to be told to.)<br>><br>> One more observation, about a specific time: Last night's out-of-sync
<br>> break occurred at 36:19 into the Charlie Rose show. That's within the<br>> first two minutes of Letterman. Maybe there is extra overhead at the start<br>> of a recording?<br><br>Possibly.</blockquote><div>
<br>
If there is a problem with starting a second recording when a first recording<br>
is underway, then it should be repeatable. I may set up some tests that<br>
run during Daytime TV, to see if I can get the out-of-sync problem to<br>
repeat, reliably.<br>
<br>
-- <br>
MM<br>
<br>
<br>
</div></div>