Power problems

[Lord Albion]

Hello all,

I am currently running Kubuntu 7.10 with an Intel Core 2 Duo (E series chip) running on a Gigabyte motherboard, 2GB DDRII Ram, Gigabyte graphics card, two HDD's and a DVD drive. I cannot remember the exact details of these items as i am at work and the manual's are at home. I have been having problems with the system as the power keeps on cutting out, sometimes 5 times an hour or even more but sometimes it will be fine for days. I have replaced all the hardware so far, except for the drives and the fault is exactly the same. I took it to a computer shop on Saturday for them to take a look but they couldn't find anything wrong with the hardware.

Does anyone know what might be wrong?

Regards,

A.

[dibl]

Re: Power problems

I once saw that kind of trouble caused by a "cold" solder joint in the PSU -- it would always start and boot, but once it got hot, it lost contact and rebooted the system.

I would suspect something like that, or maybe a thermal issue with the CPU -- the new ones are designed to shut off before damage occurs when they get hot. If you install the lm-sensors package, and run "sensors-detect", then restart the system, Ksysguard should pick up the CPU thermal sensors. Gkrellm does, for sure. HTH

[Qqmike]

Re: Power problems

what dibl said

You said you replaced the h/w -- that includes getting a known-good, quality power supply (not some white-box, generic Pacific rim pos)?

If it's heat related, check (1) the proper seating of the HSF unit on the CPU, and (2) make sure your heatsink fan unit is working and attached properly (eg, plugged in to the right power header).

[Lord Albion]

Re: Power problems

Thanks. I replaced the PSU with a unit I know works well. I also replaced the processor, graphics card, memory and CPU cooler (I got the biggest heatsink and fan I could find). The only remaining parts are the two hard drives and DVD drive. The shop I took it two said they stress tested the unit and installed Win XP on it over the weekend with no problems at all. They also ran diagnostics on it and came up with no errors.

[dibl]

Re: Power problems

I'd say install lm-sensors, hddtemp, and gkrellm, via Adept Manager. Once installed, open the Konsole and enter

sudo sensors-detect

and follow the script. When you get to the question about writing the modification to /etc/module, you have to type "yes" because the default is "NO". Then restart your system.

After the restart, do Alt-F2 and enter "gkrellm" with no quote marks. Right-click on the top of the panel to configure it, or follow the prompt when you first run it. On "Sensors" click "thermal" and put a check mark in the CPU(s) and each hard drive. I also widen the Gkrellm panel so there's a decent time interval displayed.

Then you can watch and see if something is getting hot, or if we're totally off base with the thermal theory.

[Lord Albion]

Re: Power problems

They are installed now, many thanks.

The temperature readings seem fine.

Core 1 = 30.0C
Core 2 = 29.0C
HDD 1 = 36.0C
HDD 2 = 34.0C

All fans are also running well (PSU, CPU and both case ones). I shall leave the machine on overnight and tomorrow during work so if it goes off I can review the logs.

[dibl]

Re: Power problems

Core 1 = 30.0C
Core 2 = 29.0C
HDD 1 = 36.0C
HDD 2 = 34.0C

Wow, I wish my Core 2 Extreme would run that cool!

If it turns out not to be a thermal issue (and of course, your PSU still isn't instrumented ....), then I'm kinda outta gas on this one. Yes, look at the logs, but unless it's some goofy power-saving conflict, it's hard for me to imagine software being the cause of intermittent power-downs. :P

[Lord Albion]

Re: Power problems

I paid extra for a nice heatsink/fan, even though the machine is not as silent as claimed :P

I doubt it is a power saving setting as it has happened at random intervals, even during installation (several times). I was running 7.10 on my old hardware but had to reinstall it because of the massive hardware change due to drivers, I don't know how to install drivers manually yet.

[Qqmike]

Re: Power problems

memTest? Run it for 4-8 hours.
Faults in the memory can sometimes produce such symptoms.

And, as the saying goes, It's always a cable!
Might replace power (and data?) cables connecting to mobo headers.

[Lord Albion]

Re: Power problems

Thanks for that but I have already run memtest many times, (multiple sets) and all cables in the machine are new, apart from the data cables for the hard drives.

The thing I don't like about memtest is that it doesn't allow one to test all the memory (as far as I understand it) as there is always the area of memory used by the computer to run the program, even if it is not a large area I would still prefer to be able to test all of it.

I think it would be a good idea for Kubunty to come with a stress-testing or general diagnostic package instead of just memtest so we can test processors, motherboards, etc.

[moir]

Re: Power problems

Just install mprime, run 'mprime -m' and set up for stress testing. It's the best diagnostics out there for CPU on any platform, and second only to memtest86 for RAM.

Like benchmarking, there is no perfect system test. I wouldn't expect a shutdown as a normal failure mode with a RAM problem though. Overheat, bad power, bad power supply: yes. Don't overlook a flaky power supply cord or something dumb like that.

[Lord Albion]

Re: Power problems

Thanks for that, if you meant Mersenne Prime by mprime I have already tried running SETI@Home for stress testing and the system falls over within a few minutes leaving me with no alternative other than to reinstall the OS.

I have just ordered a new hard drive as I have replaced everything in the system apart from the hard drives and DVD drive. I have already replaced all power cables, leads along with the motherboard, processor, memory, graphics card and case. I don't think it would be the DVD drive as that was left in over the weekend when it was tested. I took the computer to a local computer shop to test for me at the weekend and told them I have Linux installed on the smaller of the two HDD's which they could wipe off if they wanted for purposes of testing (they only know Windows) but that I didn't want them to touch the larger drive as that has my work on it. They disconnected both drives and used one of their own and said the system was fine all weekend until they turned it off on Monday. This leads me to suspect (hope) it may be one of the hard drives. Memtest has been left running for ages and not found any problems with the memory and the computer place said they tested everything apart the hard drives thoroughly and are satisfied they are all working fine.

Could it be something in Kubuntu 7.10 that is causing the problem?

[Lord Albion]

Re: Power problems

Hello there and thanks for your help thus far.

I have been experimenting with different distributions (Fedora Core 8 and OpenSuse 10.3). The problem seems to be with Kubuntu as neither of the others had the same problem and I left my computer on for an entire day (early morning to late at night) several days in a row. I do however want to stick to using Kubuntu as I think it is better than those two.

I am not an expert but three does seem to be something unique to Kubuntu/Ubuntu in this respect, could it be ACPI or something else to do with the power systems?

Thanks in advance.

A.