Two system hard drive crashes in two months. Is it my system setup?

finne · 2009-10-25 07:53:44

Before I buy another hard drive I'd like to know what's casing the crashes. I've installed the system on an old IDE drive now, the crashed ones were SATA.
I do get some non-continuous messages when running a hard drive check. I haven't cared much about them until the hard drives started to crash. Maybe I should? I remember having 2,3% on the first and 0,3% on the second. I am a bit surprised about the second one because it was brand new and crashed after only a month. The first I had for about a year but come to think of it, I didn't install the system on it until maybe this summer. What causes these non-continuities?

Another thing is that I don't have a swap partition. I don't see any reason to because I hardly ever reach even 10% of ram usage and I don't use a laptop with hibernate or whatever it's called.

I don't know if it's the hardware in my computer that causes the crashes or my arch setup. Any ideas?

Giacomo Ritucci · 2009-10-25 08:08:54

Non continuous is not a symptom on incoming failure, it's just a measure of how much data is scattered around your disk [1], and swap is totally unrelated.

This must be a harware problem.

[1] http://en.wikipedia.org/wiki/File_system_fragmentation

gog · 2009-10-25 08:10:02

look at hdparm and smart in the wiki

smartd will give you a lot of data on your hdd's life expectancy/reliability

Last edited by gog (2009-10-25 08:10:15)

finne · 2009-10-25 08:39:52

Thanks guys.

Any idea on how to check if it's the motherboard or the PSU that's broken? Maybe borrow a PSU and check hard drive with smart or hdparm?

gog · 2009-10-25 10:14:09

hdparm isnt a monitoring program. it sets hd parameters

after youve read about both of them in the wiki, you could use smartd to tell you how many times the hdd spins down per hour. if its too much you could use hdparm to turn off or decrement power management on the hdd in order to preserve its life

this is just an example of what you can do with the 2. in your case, youre probably more interested in monitoring so smartd is all you need (for the hdd). i only mentioned hdparm since theyre usually used in companion

you could see the different effects 2+ psus have on your hdd, but its awkward with smartd, as you cant automatically make it parse results from past xyz hours; it will list values recorded since the first time the drive got powered on.

since it has values on "time powered on" and spin down, bad sector count, head parking, etc. you could divide those and get what you want. there is a lot of data and you can get an idea of what to look for by reading this thread: http://bbs.archlinux.org/viewtopic.php?id=39258

e: about diagnosing the psu: i have no clue about this one, and i dont recall seeing this covered in the wiki. if you do find some sites on this for linux, let me know

Last edited by gog (2009-10-25 10:20:40)

deej · 2009-10-25 10:35:29

You might also check the hard drives' manufacturers web-site for software to properly set-up
your SATA drive. I recall that mine [ Hitachi SATA II ] required a little tweaking when I installed
it. If memory serves, these drives ship with a 'general' setting as default - mine was as noisy as
a couple of Californians on a Vespa until I downloaded the software and tweaked it.

Deej

aniruddha · 2009-10-25 18:12:57

finne wrote:

Before I buy another hard drive I'd like to know what's casing the crashes. I've installed the system on an old IDE drive now, the crashed ones were SATA.
I do get some non-continuous messages when running a hard drive check. I haven't cared much about them until the hard drives started to crash.

What do you mean with crashes? Do the hard drives still spin? Can you boot, do they make a ticking sound?

Another thing is that I don't have a swap partition. I don't see any reason to because I hardly ever reach even 10% of ram usage and I don't use a laptop with hibernate or whatever it's called.

You should alway use a swap partition, but that's another discussion

Any idea on how to check if it's the motherboard or the PSU that's broken? Maybe borrow a PSU and check hard drive with smart or hdparm?

You can use another psu or use a psu tester: http://www.legitreviews.com/article/267/3/

finne · 2009-11-03 07:53:54

Update.
I've tested the psu with a psu tester. There were a few values that weren't okay. 5V was 5,4V and one of the 12V values was under 10V. I don't really know what this means. I probably need to change the psu but I don't know if it broke the hard drives.

I'll listen at one of the hard drives (I took out the magnets from one of them, they're fun!) when I get home from work tonight.

After the first crash I remember being able to get the hard drive going again, after a format, but it broke soon after again. I haven't done anything with the second hard drive since it broke.

Edit:
I've had ide disks (the broken ones are sata) connected to the same psu for several years without crashing. These disks have been connected through a controller card. Maybe it is the motherboard after all?

Last edited by finne (2009-11-03 09:46:24)

zenlord · 2009-11-03 11:56:26

Those kind of problems are very hard to diagnose. It could be either the PSU, either the motherboard or both. Plus it could be even bad luck with a flawed hard drive.

Check your motherboard for (almost) broken capacitators: look for little turrets with a cross on top. If the top is open, then it is broken. If the top is not flat, then it is just a matter of time before it'll break.

Check your PSU readings. It's not clear when the voltages are bad: some say a difference of 5% is allowed where others say 10% difference is normal. If you tell me that your +12v gives only +10V, then either your PSU is faulty of your mobo is reading it badly.

Is the PSU a decent one - and I'm not only talking about raw power? Where I live, a decent PSU costs around €50, but you can easily find one for half that price. I never rely on cheap PSU's anymore, because they introduce instability to your system that is very hard to diagnose.

cerbie · 2009-11-03 12:48:16

finne wrote:

Update.
I've tested the psu with a psu tester. There were a few values that weren't okay. 5V was 5,4V and one of the 12V values was under 10V. I don't really know what this means. I probably need to change the psu but I don't know if it broke the hard drives.
I'll listen at one of the hard drives (I took out the magnets from one of them, they're fun!) when I get home from work tonight.
After the first crash I remember being able to get the hard drive going again, after a format, but it broke soon after again. I haven't done anything with the second hard drive since it broke.
Edit:
I've had ide disks (the broken ones are sata) connected to the same psu for several years without crashing. These disks have been connected through a controller card. Maybe it is the motherboard after all?

Can you trust that PSU tester? I like to go for the risky way and use a DMM while the system is stressed . However, if you trust those readings to be correct, RUN, don't walk, to get a better PSU. 11.4V should be the minimum +12V reading. I wouldn't be surprised if the card is helping show you problems, but again, if those readings are correct, the PSU is not performing adequately. bad PSUs can cause all kinds of strange problems, which are often difficult to diagnose (until a new PSU fixes everything).

Any 80+ PSU made by Seasonic or Delta aught to be good (Seasonic, Corsair, OCZ, Antec, etc.).

some say a difference of 5% is allowed where others say 10% difference is normal

+5% (page 22).

Last edited by cerbie (2009-11-03 12:49:11)

R00KIE · 2009-11-03 13:38:22

You may as well refer to the datasheets of the components, hard disk in this case. Those tolerances are what the atx spec says, sometimes those tolerances are not respected, most psu's (even expensive ones) can have somewhat bad regulation from what I've seen, and the output voltages do change quite a bit with load.

Of course a cheaper psu will be worse, aging does not help too, also I would try to check the quality of your service, it shouldn't matter much now with psus having active pfc and all that but theory and practice can be quite different.

finne · 2009-11-04 16:44:38

Okay, I tried reconnecting the broken sata drive and it works! There were some complaints about the file system at startup but it works. I don't trust it though. I'll have to do backups of everything now that I can...

Btw, it's a Codegen 480W psu. Don't know if it's good or not. It cost me about $50 four years ago. I bought it in GB (I live in Sweden) so maybe that's what's bothering it? Both countries use 220V though...

Well, back to the backuping...

Edit:
Both hard drives are Samsung, 320 GB and 1 TB.

Last edited by finne (2009-11-04 17:01:36)

Arch Linux

#1 2009-10-25 07:53:44

Two system hard drive crashes in two months. Is it my system setup?

#2 2009-10-25 08:08:54

Re: Two system hard drive crashes in two months. Is it my system setup?

#3 2009-10-25 08:10:02

Re: Two system hard drive crashes in two months. Is it my system setup?

#4 2009-10-25 08:39:52

Re: Two system hard drive crashes in two months. Is it my system setup?

#5 2009-10-25 10:14:09

Re: Two system hard drive crashes in two months. Is it my system setup?

#6 2009-10-25 10:35:29

Re: Two system hard drive crashes in two months. Is it my system setup?

#7 2009-10-25 18:12:57

Re: Two system hard drive crashes in two months. Is it my system setup?

#8 2009-11-03 07:53:54

Re: Two system hard drive crashes in two months. Is it my system setup?

#9 2009-11-03 11:56:26

Re: Two system hard drive crashes in two months. Is it my system setup?

#10 2009-11-03 12:48:16

Re: Two system hard drive crashes in two months. Is it my system setup?

#11 2009-11-03 13:38:22

Re: Two system hard drive crashes in two months. Is it my system setup?

#12 2009-11-04 16:44:38

Re: Two system hard drive crashes in two months. Is it my system setup?

Board footer