On 9/10/07, <b class="gmail_sendername">Taco Mel</b> <<a href="mailto:taco_mel@yahoo.com">taco_mel@yahoo.com</a>> wrote:<div><span class="gmail_quote"></span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
With my current settings I have experienced a problem<br>with playing back HDTV (prebuffer pauses, very<br>jittery) when the back end server is "busy":<br><br>1. Recording HDTV stream with pcHDTV 5500<br>2. Recording SDTV stream with Hauppauge 500
<br>3. Commercial flagging #1 and #2<br>4. Feeding HDTV stream #1 to remote front end<br><br>The back end has 1 TB storage in RAID-5 and is a dual<br>Xeon 1.8 GHz setup, hyperthreading enabled. Linux<br>sees this as 4 CPU's. There I find a "mythcommflag"
<br>process using 100% of a CPU (iowait around 5-20%).<br>The front end (Athlon X2 5200+) isn't even breaking a<br>sweat. I tried to "renice" the mythcommflag process to<br>+19 on the back end and this did not help. Killing
<br>mythcommflag on the back end immediately "fixed" the<br>problem.<br><br>While it's easy enough for me to log in to the back<br>end to kill the commercial flagging, I'd like<br>something more automated and wife-friendly than this.
<br><br>If I export the recordings directory via NFS is it<br>possible to run the commercial flagging from another<br>machine -- perhaps on the aforementioned front end<br>which has plenty of free cycles? Preferably just for
<br>this one source? Should I turn off hyperthreading on<br>the back end? Any better ideas?</blockquote><div><br>Not to point out something simple but I missed this on my recent addition of two 500GB drives to my system. Do all your drives have DMA enabled? Without, my new drives were very unusable beyond one activity. Also, if you have the ability to put your /var partition on a separate drive/array from your recordings, this makes a big improvement with all the DB read/write activity coupled with the disk activity.
<br><br>Kevin</div></div><br>