0

I have a problem similar to the one described on question Why are my hard drives failing?. I have a PC assembled by my self and It has worked fine for 3 years but recently It have been killing hard drives very quickly, a 2TB Western Digitals, a 1 TB Seagate, a 2TB Seagate. Each one lasted less than 6 months, the last was still under warranty (less than 3 months). Then I thought that there was a problem with the PC so I bought a new one. The new one killed the substitute of 2TB Seagate and I got another third hard drive from the shop under the same warranty (haven't installed it). I've been using a 320GB that came with the new PC but it is starting to fail. Some times the new computer fails (I think) because of the heat. So I bought a new super artic cooler fan for the processor and a new super special power supply. And even thought the processor's temperature have dropped significantly the problem still happens, the PC shuts down under heavy processor load.

Ok, so what is different with the earlier question (why are my hard drives failling?):

  • Both the old and the new PC always had metal computer cases, bought on a shop, not made by me.
  • I don't have any RAIDS or anything just a regular PC with one drive that spends most of the day shut down.
  • I'm 98.79% sure that my computer is correctly grounded. All the components inside it are sharing the "common ground" of the power supply.
  • I've been using debian 8 for both PCs. The old one had debian 7 but I made the upgrade.
  • As you can see I already used 3 different power supplies, the old one, the new one, the super special one.
  • The common factor between the 2 computers is the power at my home. Both were fed through an inverter. The inverter had some repairs the last year.

Anyway I'm almost sure that I'm having electric problems, the inverter, or may be my home's electric system, in this country (Dominican Republic) power could have some problems. I need suggestions on things that I should look for, but specially what are the electric parameters that I should check. I already checked the voltage an its 120V AC, with and without the inverter. Should I check the frequency?, power (amps) consumed by the PC? any other thing besides power? should I measure this parameters inside the case in the line that feeds the hard disk? Any other suggestion, measure or experiment?

I don't know if I should have opened a new question, but because my problem is a bit different I did it, so I also want to know if I did right (for future references).

UPDATE 16/07/2015 UTC 17:14:03

Thomas Frank: Thanks, I forgot that part, When the drives fail they can be seen by Debian, but I receive a lot of I/O errors "can't write this", "can't read that", they fail to boot and the Debian installer fails to partition them. I can boot through a usbstick and see the file system and copy the data to other drive. One of the death hard disks reported a 0 size on the bios. I haven't made a lot of checks to the hard drives, haven't use the gnome-disk utility to check on the smart data. I'll try to test them all later today.

UPDATE 19/07/2015 UTC 03:23:03

I'm adding the information asked by Geruta, AcePL and Frank. I've been testing a few more things so I have new info:

Geruta was kind of sharing the electric problem theory so I took the computer to my girlfriends home and test it there and it restarted. The only thing I did there to create load for the computer was running blender, in the default screen that has a camera a cube and light, I moved the 3d cursor, hit the space bar and typed "add monkey", then pressed "a" to "select all" and hit "shift + d" to duplicate all the mentioned objects, and then "select all" again and "duplicate" again. It endured 12/13 duplicates and restarted. Almost all the times when it restarts the next time it tries to boot, debian shows some hardware error message and restarts again, and again and again. I only let it do it 4 times cause I was afraid.

This is the message each time it boots up other day (https:// goo.gl /photos/G1f6gDDAQqPa3ptH9 , https:// goo.gl /photos/FQRjqXqUTQ4P48pe7

from the image (some words translated from spanish):

  • bloque = block
  • fichero = file
  • limpio = clean
  • recuperando = recovering
  • transacciones = transactions

I took the computer back to store were I bought it cause the technician there wanted to test it, and he said that nothing wrong happened. He tested it with he's own hard drive with windows and some "hardcore games", not the 320GB WD that I use with debian 8. Any way he suspects of the heat so he added a new fan to the case. He also presented me a friend that knows about inverters. We spend an hour playing with the inverter at my home and it turns out that it is a bit old (almost 10 years) and that it has some problems. During our tests some times the frequency output of the inverter was 43/47 Hz not the 60Hz that it is supposed to give. We disconnected the inverter because in our test we couldn't got it to work again, now the power comes from the normal electric system 120v 60Hz (all frequency and voltage measures made with a multimeter), but now I'm at the mercy of black-outs, so I have to repair the good old inverter.

This is the memory test photo https:// goo.gl /photos/yA8LBVypFn2Fiqy36.

Today in the morning tried to watch some videos via streaming on the new PC and it restarted. I tried to get some data from two of the damaged hard drives that are still around to answer AcePL, the 2TB WD and the 1TB SG, the late 2 2TB SGs had to be taken to the store to claim the warranty and they checked them before giving me a replacement, the 2 times, and it turns out that this two hard drives are not dead, they booted, they still have debian 7, they have all my files, and the SMART info in the gnome-disk-utility of both hard drives seems to be fine, I fill embarrassed, cause I remember that I tried to use them a few times before giving up. After the restart in the morning I've been doing several tests to try to isolate the problem but nothing I do seems to make it fail, the blender test, cpuburn from the debian repository, more streaming, some 3d games, memtest86+, combinations of the previous things and nothing. It has become very very hot at some times, the CPU as hot as 66C, hard drives too, the SMART info reported 47C, I suspect that they got hotter when I wasn't looking. But the hottest component i think is the ATI video card. The first half of the day the case was open.

Image from the disks and the PCs https:// goo.gl /photos/mqyLEoWJPssEppBv6

Old PC:

  • Board: ASRock G31M-VS2
  • CPU: Pentium Dual Core E5400
  • RAM: 1GB DDR2 333MHz, 2GB DDR2 333MHz

New PC:

  • Board: ASRock 960GM-VGS3 FX
  • CPU: AMD FX 6300 "six core"
  • RAM: 4GB DDR3 533MHz
  • Video: ATI HD 6450 1GB (fan-less, just an aluminum sink)
  • Power supply: 550W LOGISYS
  • CPU Fan: Artic Freezer 7 Pro ver. 2
  • Why do you use a cardboard box for a case? – Frank Thomas Jul 16 '15 at 17:00
  • So for a start in what way are they failing? what do the SMART stats for one of the failed disks say about why it failed? in debian you can install gnome-disks (palimsest) to read SMART stats. – Frank Thomas Jul 16 '15 at 17:02
  • jajajaja As I said before I'm using two real computer cases, but the card board box idea was really cool, I'll try that some day, for now the information you asked is added to the question. – programingfrik Jul 16 '15 at 17:33
  • How many watts is the power supply? Did you use monitoring tools to check the voltage on the other side of the power supply? (speedfan) – cybernard Jul 17 '15 at 3:58
  • The new computer power supply is branded "one" and its 700W. The power supply that I bought (the super special one) is a 550W LOGISYS. I just used a multimeter to test the voltage in the electric outlet where the computer is connected. – programingfrik Jul 18 '15 at 19:35
1

Hard drives are most visible, but just one symptom of the problem. Everything you mention points to two possibilities.

First one is overheating. You mention that PC shuts down under load with new PSU and new cooler. You mention that you're in Dominican Rep. Well, there you have it... Do you use computer with box closed or open? It should be closed, as only then there will be proper cooling to all components. With open box CPU gets cooler, but everything else gets hotter. What was the HDD that was first in the box 3 years ago? Can you tell us what was it's manufacturer and model (RPM especially).

Second possibility has to do with the manner the hard drives are failing. What is the most common failure type? SMART info would be a start. This sort of determines the cause of the problems. It may be unstable power supply from the grid... But I think it's unlikely - in that case you would see more random failures in all components, not only HDD. Did other electronic devices in the household failed lately? TV, stereo, microwave etc.? If not, then no: not power supply. I would instead see if computer gets damp inside. If it's possible that computer cools fast after power down it's more than likely that there will be moisture inside. Physics. And HDDs are not hermetically sealed, so they are susceptible to liquids.

Did both PC's stood in the same place? If yes - is the spot shaded or just lately is in the sun? Look at bits like that...

  • I think that it was a combination of the two possibilities, I changed the processor fan, and added an extra fan to the case so it was cooler and also sent the inverter to a technician and he made some calibrations. I didn't to much research after that so the true reason for the problem is not clear, but its not happening any more, thanks! – programingfrik Mar 27 '16 at 20:09
1

Hard drives will only fail that frequently for one of two reasons. Either you are not providing proper cooling or you are not providing proper power.

Cooling is normally not an issue for HDs, some it's likely your having trouble keeping a good electrical signal. You made mention that you tried a new PC. Did that new PC have a different power supply?

My recommendation: get a surge protector if you don't already have one, get a new PSU that's built to handle your power wattage usage. Check all electrical connections and ensure everything is connected securely and in the right place.

  • Actually thermal considerations are very important to HDDs but they are slow to heat up (and to cool down), and are generally situated in the case such that there isn't usually a lot of heat accumulation. That said though, they have the narrowest safe temperature range of any common component (generally about 25-45C). – Frank Thomas Jul 16 '15 at 17:13
  • Hence the reason I listed it as one of the two reasons but also noted that it's not normally the issue. Simply, because it's not. Do searches for had drives overheating and you'll see it's not that common of a problem. – Geruta Jul 16 '15 at 17:17
  • Ok, I'll make some tests with the power, a friend suggested me to take the computer to other house and test it there. I'll try that and I'll try to check the hard drives temperature, the information obtain will be added. – programingfrik Jul 16 '15 at 17:40
  • @FrankThomas it's more like 15 - 55C range, so not that bad, but you're still right that it's narrowest of 'em all and it's not normal operating range. Also, desktop models are quite good at heat dissipation - again, you're right that location helps - but all depends. Put them in SFF case and it's all gone... – AcePL Jul 16 '15 at 17:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.