LinuxMCE Forums

General => Users => Topic started by: fearingsept on July 30, 2008, 04:44:37 PM

Title: Core Locking up, Most common causes
Post by: fearingsept on July 30, 2008, 04:44:37 PM
I have had my hybrid core up for over a week now. It works pretty good except every now and then it seems to lock up. I notice it first at the MD since the hybrid is in my closet. It didn't seem to have a problem until I installed my MD. The MD works great when the system is up. I don't know for sure if the MD is the actual cause or if it could be that the core is finally controlling my network and for some reason it is crashing. Are there some common things I can check?
Title: Re: Core Locking up, Most common causes
Post by: jeangot on July 30, 2008, 07:18:28 PM
You could start by checking fans on the CPU and power supply to confirm that they're working. Lockups on any of my computers (not specific to LMCE) are usually caused by heat.
Try leaving your closet door open, and perhaps even the computer case open, and see if it makes a difference. At least that way you can rule out heat as being the problem.
Sometimes, bad memory chips can also cause lockups. If you have spare memory, try that...

I actually had an issue with my own core recently. It wasn't locking up, but rebooting. The cause turned out to be a defective/old power supply.

Good luck.

Jean
Title: Re: Core Locking up, Most common causes
Post by: Itsik on July 31, 2008, 07:23:05 AM
The main cause for lockups in lmce, providing your hardware is all compatible and setup correctly, is heat. Start at the northbridge (your graphic chipset) as it is pretty sensitive yet frequently over looked.
Once you have your cooling figured out, lockups should be a thing of the past.
Make sure your MD's are also on the cool side as an MD lockup will not crush the system but frequently will slow down (maybe even stop) your network.

Itsik
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on July 31, 2008, 07:40:25 AM
Quote from: Itsik on July 31, 2008, 07:23:05 AM
The main cause for lockups in lmce, providing your hardware is all compatible and setup correctly, is heat. Start at the northbridge (your graphic chipset) as it is pretty sensitive yet frequently over looked.
Once you have your cooling figured out, lockups should be a thing of the past.
Make sure your MD's are also on the cool side as an MD lockup will not crush the system but frequently will slow down (maybe even stop) your network.

Itsik

In our experience 'over heating' on its own is very rarely the cause of lockups. A good simple test is to boot the system that locks up under a Kubuntu Live CD and leave it up for a few hours/days to see if it locks. If it doesn't then you know that its something related to your LinuxMCE config thats causing the problem... then the real detective work begins ;-)

Andrew
Title: Re: Core Locking up, Most common causes
Post by: jeangot on July 31, 2008, 05:12:04 PM
Andrew,

if you boot the Core with a live CD and just let it sit idle, wouldn't that place a lot less load on the system than a live working core, hence produce much less heat?
So it wouldn't really test for heat, but it would test the hardware by itself (memory corruption etc) ?
If it wouldn't lock with the live CD, it could still in theory be due to heat, which would be much higher on the working core config (in my opinion).

Jean
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on July 31, 2008, 05:36:55 PM
Quote from: jeangot on July 31, 2008, 05:12:04 PM
Andrew,

if you boot the Core with a live CD and just let it sit idle, wouldn't that place a lot less load on the system than a live working core, hence produce much less heat?
So it wouldn't really test for heat, but it would test the hardware by itself (memory corruption etc) ?
If it wouldn't lock with the live CD, it could still in theory be due to heat, which would be much higher on the working core config (in my opinion).

Jean

Well in our experience if you have any kind of overheating problem (which really is incredibly rare) then its mainly down to badly attached/seated heat sinks or possibly failed or failing fan assemblies. However in either of those cases the overheating will happen even at idle.

Running from a Live CD or possibly running MemTest will in almost all cases allow a failure to happen.

Andrew
Title: Re: Core Locking up, Most common causes
Post by: Zaerc on July 31, 2008, 07:46:27 PM
I think you're forgetting that a lot of people are trying to cram hot machines in tight/closed spaces/cupboards, perefeably without proper fans as they would make to much noise... 

Now without proper air flow most machines will overheat sooner or later, I have seen this on many occasions.
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on July 31, 2008, 08:29:35 PM
Quote from: Zaerc on July 31, 2008, 07:46:27 PM
I think you're forgetting that a lot of people are trying to cram hot machines in tight/closed spaces/cupboards, perefeably without proper fans as they would make to much noise... 

Now without proper air flow most machines will overheat sooner or later, I have seen this on many occasions.


True... but in that case the system will overheat for sure from a LiveCD too... the point I think is to eliminate LiinuxMCE from the loop.

Andrew
Title: Re: Core Locking up, Most common causes
Post by: fearingsept on July 31, 2008, 08:31:33 PM
Ok.. I May have tracked down the problem. Thank you to everyone on this one. I do believe for now that It is a Heat issue on my core. And not a normal one even...
So I bought a Rosewill NIC that has an "Optional" Heat sink. This card goes from my core to my switch which as we all know interfaces with the MD's and other systems on my network for internet use.
http://www.newegg.com/Product/Product.aspx?Item=N82E16833166002

Well, I left the heat sink off when I installed the card. I had never had a NIC that needed a heat sink so I figured no big deal. What I didn't think about was the fact that it will constantly be in use because of the MD and other systems on my network. I believe that my card might be over heating. My whole system doesn't crash to a point of shutdown.. It locks up and it seems to lock up the network before anything else.
I have 2 more of the same card with the heat sink already on it. I may just swap out just in case there was any damage done to the card in question.
I will keep everyone posted on if this actually fixes my problem.
Title: Re: Core Locking up, Most common causes
Post by: jeangot on August 01, 2008, 12:39:37 AM
Zaerc: very true, in several places I went into my lower cabinets and cut a hole in the fake bottom of the cabinets, to hide my MDs underneath. Of course they wouldn't last very long because there is hardly any ventilation, and the hotter they get, the more the fan runs, which makes the MDs even louder... Haven't found any affordable/powerfull fanless solution yet...

Jean
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on August 01, 2008, 12:58:22 AM
Quote from: jeangot on August 01, 2008, 12:39:37 AM
Zaerc: very true, in several places I went into my lower cabinets and cut a hole in the fake bottom of the cabinets, to hide my MDs underneath. Of course they wouldn't last very long because there is hardly any ventilation, and the hotter they get, the more the fan runs, which makes the MDs even louder... Haven't found any affordable/powerfull fanless solution yet...

Jean

So another simple test would be to take your Core and MD's out of whatever cabinets they are in temporarily ie provide them with plenty of ventilation... and see if they still lock up.

On the question of fanless units we are testing new Intel Atom 230 (i945 based with GMA950 GPU's) based MD designs now that are totally silent and work at resolutions up to 720p currently.

Andrew
Title: Re: Core Locking up, Most common causes
Post by: jeangot on August 01, 2008, 01:08:49 AM
Andrew,

fearingsept is the one with the problem, I'm not complaining, I know that if those die or lockup, it will be my fault :-)

Do you have a model number for the motherboard, and a confirmed fitting case recommendation (as small as possible, if possible vesa mount), so that I can start playing with it over here? How are those different from the Fiire Invisible units that seeme dto do only 720p?
have you every tried a Jetway JNC62K? It is a Mini ITX board with onboard Nvidia and AMD Cpu. The problem is I'm not sure if I can run an AMD processor without any fan, even a Brisbane that runs cooler.

(Sorry for getting off topic)

Jean
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on August 01, 2008, 01:16:42 AM
Quote from: jeangot on August 01, 2008, 01:08:49 AM
Andrew,

fearingsept is the one with the problem, I'm not complaining, I know that if those die or lockup, it will be my fault :-)

Do you have a model number for the motherboard, and a confirmed fitting case recommendation (as small as possible, if possible vesa mount), so that I can start playing with it over here? How are those different from the Fiire Invisible units that seeme dto do only 720p?
have you every tried a Jetway JNC62K? It is a Mini ITX board with onboard Nvidia and AMD Cpu. The problem is I'm not sure if I can run an AMD processor without any fan, even a Brisbane that runs cooler.

(Sorry for getting off topic)

Jean

As far as i know the Fiire Invisible units are based on nVidia GPU's and do have fan based cooling.

Our new unit is based on the Intel D945GCLF mini-itx main board and a custom fanless heatsink & Integrated enclosure. However the Intel D945GCLF motherboard could be used with other 'off the shelf' heat pipe or passive heatsink based cooling and an off the shelf case.

Never tried the Jetway you mentioned for exactly the reasons you also mention!

Andrew
Title: Re: Core Locking up, Most common causes
Post by: jeangot on August 01, 2008, 01:36:27 AM
Andrew,

$75 bucks for the motherboard with the CPU? WOW, that's a great price to build a fanless 720p MD, but how exactly do you solve the fact that there is no SPDIF out? Add a low profile sound card? Or are there people who will use this to play video and use analogue sound?
And how do you get high def video without even a DVI port? Are you adding a separate video card? Then there is no more slot for the sound card?
Maybe I'm looking at the wrong mobo: http://www.intel.com/Products/Desktop/Motherboards/D945GCLF/D945GCLF-overview.htm

Glad we're on the same page about the Jetway board, I was wondering if I had missed a workable solution here...

Jean
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on August 01, 2008, 07:21:43 AM
Quote from: jeangot on August 01, 2008, 01:36:27 AM
Andrew,

$75 bucks for the motherboard with the CPU? WOW, that's a great price to build a fanless 720p MD, but how exactly do you solve the fact that there is no SPDIF out? Add a low profile sound card? Or are there people who will use this to play video and use analogue sound?
And how do you get high def video without even a DVI port? Are you adding a separate video card? Then there is no more slot for the sound card?
Maybe I'm looking at the wrong mobo: http://www.intel.com/Products/Desktop/Motherboards/D945GCLF/D945GCLF-overview.htm

Glad we're on the same page about the Jetway board, I was wondering if I had missed a workable solution here...

Jean

SpDIF can be routed out of the Audio jacks as on other mothboards. We currently use this motherboard with Flat Panel screens that have VGA inputs and drive them at 720p but we do expect their to be an option on this motherboard for an ADD card that will deliver a DVI out. There are also alternative ATOM based motherboards coming to market from all the expected manufacturers that will fix all these slight limitiations with the Intel board pretty quickly.

Andrew
Title: Re: Core Locking up, Most common causes
Post by: jeangot on August 01, 2008, 07:40:26 AM
Andrew,

thanks for the info, I had seen that you could get SPDIF out of one audio jack on the Jetway board, but didn't know that you could do this with all boards.
I use a lot of inexpensive 19'' Monitors in my house as secondary TVs, and there the VGA port and analog audio would work just fine for me. I suppose I could also try using a PCI Nvidia 6200 low profile if it would fit in the case?
If you tune to a HDTV channel that broadcasts in 1080p, is the box powerfull enough to downscale that and play it in 720p?

Thanks again for sharing all your secrets with us!

Jean
Title: Re: Core Locking up, Most common causes
Post by: totallymaxed on August 01, 2008, 10:56:07 AM
Quote from: jeangot on August 01, 2008, 07:40:26 AM
Andrew,

thanks for the info, I had seen that you could get SPDIF out of one audio jack on the Jetway board, but didn't know that you could do this with all boards.
I use a lot of inexpensive 19'' Monitors in my house as secondary TVs, and there the VGA port and analog audio would work just fine for me. I suppose I could also try using a PCI Nvidia 6200 low profile if it would fit in the case?
If you tune to a HDTV channel that broadcasts in 1080p, is the box powerfull enough to downscale that and play it in 720p?

Thanks again for sharing all your secrets with us!

Jean

Well not all boards can route spdif out of the audio jacks. As for downscaling from 1080p to 720p... we have not tested that at all as its not really an issue here in the UK. Currently the D945GCLF is driving 720P screens very nicely but 1080p is not working yet.

All the best

Andrew
Title: Re: Core Locking up, Most common causes
Post by: amirul on August 01, 2008, 05:25:18 PM
Andrew,
I see that the Intel Atom has arrived. Are you running the screen saver static to avoid locking up, or have you found a fix?

Amirul
Title: Re: Core Locking up, Most common causes
Post by: fearingsept on August 02, 2008, 04:17:40 AM
quick question...
Do I need to do anything specific when I remove my old Network card and replace it with an identical one. I just tried and well... I had to put the old one back in to get my network back up....
Title: Re: Core Locking up, Most common causes
Post by: colinjones on August 02, 2008, 05:56:55 AM
dunno, just a thought, but perhaps do a lspci -vnn before and after changing it, then compare the PCI bus IDs and vendor:product id's, rev number and maybe some of the other parameters? I would have thought if those are all the same the kernel shouldn't be able to tell the difference...

only other thing could be the way the kernel assigns the ethx names during start up. If they aren't exactly eth0 for the external LAN and eth1 for the Internal then LMCE won't be able to use them properly. Use ethtool to compare the MAC addresses with the IP addresses to make sure the right subnet is connected to the right physical cable, and that the names are 0 and 1 and the right way around. They maybe swapped or even eth0 and eth2 or eth1 and eth2.... if so you can use a udev rule to force them back....
Title: Re: Core Locking up, Most common causes
Post by: fearingsept on August 05, 2008, 08:04:17 PM
I have come to the conclusion that I do not have a heat problem with my core. If for some reason my MD was overheating or crashing would it lock up my network and stall out the core?
Title: Re: Core Locking up, Most common causes
Post by: fearingsept on August 08, 2008, 04:24:19 PM
Ok... So I swapped out my MD with a Machine that I was going to use as a second MD. So far no network issues. It seems that for some reason or another my MD was slowing down and then eventually bringing down my core. I also noticed that with my new MD my video tearing issue is gone. I pulled the video card and the capture card from the first MD and installed them on the new one. I was waiting to buy a new video card so when I took the old MD off-line I moved the video card over so I could use the new one.
So far so good. Hopefully it stays that way.