Please help with an unstable DIY NAS

[yasu]

Hello,

I have an unstable DIY NAS on my hands and it's beyond my abilities to diagnose. Any tips would be highly appreciated. It's exhibiting sporadic and unpredictable behavior, but I'll do my best to summarize as succinctly as possible.

Here's the config (mostly lifted from another SNB forum poster):

• Chenbro ES34069
• JetWay JNC81-LF
• AMD Athlon X2 4850e 2.5GHz Socket AM2 45W
• Kingston 4GB (2 x 2GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400)
• RAID5: 4x Western Digital Caviar Green WD10EADS 1TB
• System drive: Western Digital Scorpio WD800BEVE 80GB
• Panasonic SR-8178-B Slimline Tray-loading CD/DVD-ROM

- First off, I should mention that it ran fine for a week after initial construction while I installed Ubuntu Server, installed packages, modified config files, etc.

- Then I tried to kick off a Handbrake video encode to stress test the CPU; this made the system completely non-responsive after about 20 minutes

- Now, during various resets, cold boots, and yanking of non-essential bits off hardware, there was a lot of random behavior which can be summed up as: freezes arbitrarily. Most commonly, it won't boot at all: no beep, no bios output. It seems like yanking the IDE cable from the MOBO will often get it to at least show the BIOS again (but not necessarily make it to the OS). Which might make you think it's the system HDD or the DVD, but it's not, because with neither of those connected, but with the IDE cable still in the MOBO, it will still not boot. Only once it's yanked from the MOBO itself will it do something.

- Sometimes the freeze will happen later, like partway through the BIOS check or while Linux is starting, or even just sitting at the Ubuntu command prompt doing nothing. But once it has crashed once, it will usually freeze sometime during or before the BIOS load from then on.

- The day after the first crash, it ran fine all day long and I used the day to test out every possible RAM chip configuration with memtest86 and several successful(!) HandBrake encodes, so I don't think it's the RAM.

- On this successful day, I had the case wide open, so I thought it might be temperature related, so I added an intake fan and have monitored the temps more closely, but there doesn't seem to be any correlation.

So, that seems to leave MOBO, CPU, and power supply. I don't know how to test any of those things without buying unnecessary duplicates. MOBO is probably my best guess because of the IDE thing, but I'd hate to RMA it and be wrong.

Any ideas?

Sorry for the long post! Thanks!!

 

[griffin518]

First post here... was reading an article and saw your post for help... hopefully I may be of some assistance.

Typically I've seen power supply issue manifest themselves in ways other than what you've described, so it's my opinion that power may be ruled out (but I could be wrong).

It sounds like either a CPU heat/defect issue or a bad motherboard (slightly less likely).

Have you installed lm-sensors? It's a handy command line utility which will show you some great information about core and ambient temperatures as well as fan speeds and voltage. Always take the cheap and easy route before replacing equipment, right? This should also give you a good idea of where your power connections stand as well... your readings really shouldn't deviate all that far from their stated voltages otherwise it might very well be your PS.

Here's a link to install information: http://ubuntuforums.org/showthread.php?t=2780

If your CPU temperatures are abnormally high (60deg C or higher) my next question would be what thermal interface material did you use between the CPU and heatsink? You wouldn't believe it but I see some manufacturers and people using the craziest stuff.

As a last ditch effort you might try installing another distro, like Fedora just to see if it's a support issue. But I doubt it. If nothing pans out I'd send the mobo back, that'll force you to re-seat the CPU in another mobo with a new application of TIM.

Good luck!
Kalle

 

[corndog]

I'll second that.

In years of helping people troubleshoot stuff like this, it has been the same nearly every single time. If you have freeze-ups at unpredictable points early in startup, it's almost ALWAYS an overheating issue.

Pull that fan off and apply some nice high quality thermal compound between the cpu and the heat sink, making sure it's a nice even smooth application. It's nearly guaranteed your problem will go away. You wouldn't believe the difference it can make on the CPU temperature.

If that doesn't help, time for a warranty replacement mobo.

 

[yasu]

Thanks for advice! I cleaned up the CPU and heatsink and reapplied the arctic silver and that did, in fact, make a difference. On the upside, the behavior is now 100% consistent. On the downside, the behavior is that there's no beep, no bios, no boot. Argh. I guess I'll try to RMA the motherboard and see how that goes.

Thanks!

 

[00Roush]

I had an issue like this with a motherboard. Turns out it was the power supply. Not sure exactly why but when I finally swapped out to a Antec Truepower 430W it works just fine every time. Before that I had problems trying to get it to boot and also resume from standby. These are problems I had since day one with the motherboard. Usually I was able to power everything down then reset the bios settings and it would power up.

If you have a good power supply in another machine you might try it. Could be the power supply in the Chenbro case is not providing enough power on one of the power rails. Just a thought.

00Roush

 

[nanook]

i had an issue like this once, i finally chased it down to the video card. the fan gummed up with dust and quit working, causing it to overheat and lockup.

on a good day, i could get a couple of hours out of it. on a normal day, i might make it to the windows screen, on a bad day, it never got past the BIOS screen.

Now, with the new video card, it's been up and running for almost a month with no reboot.