After disabling bootsplash to quash v86d burning up my system, everything ran fine for months. Then, without any changes in my system, I noticed the fans running very loudly and checked top. Both front- and backend servers were at ~100% CPU, causing the load to be in the 2-3 range. This was with no playback in progress, no programs being recorded, no commflag jobs, etc.

I searched the web and came up with a user who said UPnP was the culprit on his system. I was able to disable it on the backend following his instructions, but not the frontend because the architecture is different and the startup happens differently. After disabling UPnP, I restarted the backend. It went back to low percentages while idle, but frontend stayed at 100%. When I restart frontend it will be low CPU for a few minutes, then climb right back up to 100%. I tried forcing Qt instead of OpenGL, no luck. The frontend logs don't seem to be showing anything interesting happening. Lots of weather updates over the hours but little else.

Is there any other way to determine what is causing this high CPU usage? If it is in fact UPnP, how do I disable it on the frontend?

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

in webmin there is a more user friendly view of top under "system" called "running processes" on the top row you can sort by CPU.

_________________
My System

Do you think it's going to show me more detailed information than top which is sorted by CPU and shows mythfrontend at the top with 100% CPU? I don't think I have webmin setup and don't want to if it's going to show me the same information in a different format. If it can help diagnose this problem I'll definitely do it.

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

nope probably not
It looks alot like top
I don't think there is UPNP in the frontend as it is just sharing files on the network.
I'm assuming you don't have the screensaver turned on.
Other than playing videos I can't think of anything labor intensive that MythFrontend does.
Is it possible that the CPU is "spinning down" to a slower clock during idol time and not "spinning" back up?
Grasping at straws because I'm not a linux expert.

_________________
My System

I'm still having problems with Mythfrontend spinning up to 100% CPU while the system is doing almost nothing. I can restart it, and the process will be back up to 100% within a few minutes, even if I'm not using the menus or trying to watch programs.

Tonight the machine started locking up when my wife tried to watch a program. She was watching one pre-recorded SD program while another commflagged and a 3rd HD program was recorded on the HD Homerun. Nothing that has taxed the machine before. When I brought up top, mythfrontend was at 100% as usual, but the load was at 9.00 and climbing! It got up to 18 while I tried stopping the frontend server, killing the commflag job and stopping the recording. The load kept climbing and climbing, but mythfrontend was the only entry in top with a high CPU usage at 100%. 18 means the machine was working at 1800% capacity, right?

Is there anything else I can do to troubleshoot? It seems like programs keep coming out of nowhere to overtax the computer even when I didn't make any changes to the system. First folding@home took so much CPU I had to stop it (after 2+ years with no problem on the machine), then v86d/bootsplash, then mythbackend & mythfrontend. Is my hardware getting ready to die?

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

I would take a look at look at the health and maintenance page for that system and see if you have anything unusual showing up with the memory or disk tests, if those look ok, then I would look into the logs to see if any errors are showing up. I would focus on any logs that are unusually large.

If you are not doing so already, it would be a good practice to copy a recent backup to another machine or to a USB drive, just in case your hardware is starting to die on you.

HTH
Bruce.

I can only think of two choices for testing
1. Test hardware. It might be possible that the CPU is dying. What kind of CPU fan and heatsink do you have on the CPU? Is it possible that the CPU isn't cooling (whether the thermal grease dried up or if the fan died or whatever). If the CPU starts to overheat it will throttle itself back on speed to keep from burning up.
2. Test software try running another live version of Linux and see how it handles that. If you have a spare hard drive lying around unplug the other hard drives and install a fresh copy of LinHES on to it. Don't import anything just set up the recorders and make some dummy recordings and see what happens.

Which "paint engine" are you running OpenGL or QT? QT seems to be a little less overhead. go to Service menu->MythTV Settings->Setup->Appearance->First line.

_________________
My System

I just had my mythfrontend pegged at 106% (I don't know how either). I was toying with some m3u8 video streams and one of them hung. I added the sources as an "iptv tuner" is it possible that you might have this same problem? I had to kill it with top to make my mythtv useful again.

_________________
My System

Here's a follow-up:

I upgraded my system to 8.2 via the script. Everything went well. After rebooting, mythfrontend and mythbackend processes were solid at 100% when the system was idle (in fact, according to the GUI it was in standby).

Once again I disabled upnp by passing the "--noupnp" option in /etc/sv/mythbackend/run. After I restarted the backend with ALT+3, the backend process dropped to 0.3% CPU immediately with no other changes being made. I think that makes it clear upnp or a related process is directly causing the high CPU load on mythbackend.

I'm still unable to find a way to disable upnp in the frontend to see if that is also causing me problems there. We don't start mythfrontend the way it was started in the blog post I mentioned earlier (which I can't find today). Since it's so clearly involved in the backend and I don't need upnp, I'd like to try disabling it in mythfrontend to rule it out.

Bottom Line: How can I disable upnp in mythfrontend?

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

Update to the update:

I looked at the logs before and didn't think much of these entries in mythmaster_frontend.log:



They were appearing over and over. Since I don't use mythweather I tried disabling it in settings and restarted the frontend. I've been running about 3 minutes now and mythfrontend is still at 0.7% CPU, where normally it gets up to 100% within a minute. I might have found my frontend problem.

Obviously it would be ideal to have everything work, and these might just be symptoms of my aging hardware, but for now I'm happy for my load to stop sitting at 1.0 or even 2.0 24/7/365.

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

Going out on limb with this one, I'd say check your bios, specially check the date and time. If it's wrong change it to the actual time and date. Keep in mind the battery inside your computer could also be dying/dead. Replace it and see if that helps.

Just a hunch....

I haven't had a chance to try replacing the battery, but the BIOS clock looks good. The success I wrote about above was short-lived, because the frontend CPU went back up to 100%. In the meantime, I finally managed to figure out where to put mythfrontend arguments and was able to add the --noupnp option. The system has been going for 6 hours now and mythfrontend is sitting at 0.5%. I'll wait a day or two to be sure, but I think disabling upnp helped my machine's load drop from a 2.0 average down to about 0.2.

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

Almost a full week later, I'm still getting:


If I needed upnp I might be concerned about finding and fixing the problem, but I'm happy and WAF is high. Until the next hard drive fails, which I know is overdue for me

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM

Thanks doomedTX, I also had this issue with backend and frontend using 100% cpu when sitting doing nothing.

I disabled upnp on the backend in /etc/sv/mythbackend/run
and the frontend in /usr/LH/bin/mythfrontend-start
by adding the --noupnp option to each one.

Now running alot quieter and <1% cpu when idling

Just in case this help anyone else, /usr/LH/bin/mythfrontend-start didn't fix it for me. As far as I can tell, that file isn't used to start mythfrontend, strange as it sounds. I had to edit /etc/X11/Sessions/LinHES and put the --noupnp option in the MYTH_ARGS statement. After restarting the frontend service, I could see the "--noupnp" parameter in frontend's line in top.

I'm writing this today because I updated myth a few days ago, which apparently overwrote /etc/X11/Sessions/LinHES. I could kick myself for writing "I found it" above without documenting HOW I found it

_________________
LinHES R8.6 BE/FE GIGABYTE GA-X48-DS5 / EVGA GeForce 9800 GTX+ / 4GB DDR2 RAM