Ubuntu May Be Killing Your Laptop's Hard Drive

wwrmn writes "There's a debate going on over at bugs.launchpad.net on whether it's the Ubuntu, BIOS, hard-drive manufacturer, or pick-any-player's fault, but Ubuntu (and perhaps any OS) may be dramatically shortening the life of your laptop's hard drive due to an aggressive power-saving feature / acpi bug / OS configuration. Regardless of where the fault lies or how it's fixed, you might want to take some actions now to try to prevent the damage."

So what's new?

[TheBrutalTruth]

OpenSuse 10.2 does to, without some tweaking. I'd wager other distros have the same or similar issue. I love *nix, but it it not ready for primetime yet, with bugs like that. I shudder to think of the call from my old man, where I have to explain that he has to rebuild (like he could of in the first place) his PC with pci=nomsi and acpi=forceirqpoll in the boot options so his high dollar toy isn't ruined.

Effects on Dell/Ubuntu OEM?

[kusanagi374]

I wonder what is the effect of this bug on officially endorsed and supported Dell notebooks with Ubuntu on them? Wouldn't something like this be caught up by Dell's QA? Or is it exclusive to 7.10?

Re:AHA! :D

[Killjoy_NL]

Heh,actually, buying Windows Vista solved a problem I'd been having, random reboots during WoW and only WoW under xp. No blue screens nothing, just poof death and hey look, I have 2gb of ram wheeeeeeeeee.

Well, after buying Vista (was only +/- 18 euros since I work for an educational institution) I got a reboot too, after a nice blue screen. Best part about Vista for me is, it shows the info from the blue screen next time you boot up. I googled the error, did a little digging and it turns out 1 bios setting was wonky.
The one that clocks the cpu back if it thinks you don't need performance.
After turning that off, I never had the problem to this day.

So in all honesty, Vista works fine for me (I know it doesn't for others) and it was dirt cheap to boot :)

Re:Selected Excerpts

[tchuladdiass]

The main problem is a combination of the short spindown time, and something wanting to write out to the drive every 30 seconds or so. The main culprit could be the fact that by default, a files last access time (atime) gets updated on every read, even if that read comes from cache. So when the drive is spun down, it gets spun up even on cached reads (to write out the atime).
Add "-o noatime" to the filesystems in /etc/fstab, and that should clear up the issue.

Maybe this explains

[ILongForDarkness]

All the bitching around my work about how hard drives used to last longer. With my limited cross section, I have 2 computers at home, both ca 1998, still running original hard drives, in fact I've obsoleted 6 workstations so far at home, none of them had hard drive failures, I had one PSU, one GPU, and one NIC failure. At work (mainly a IBM shop) I've had to replace about 20% of drives within 4 years (I admin 50 workstations). I realize there is a lot of variables, smaller read heads, faster spin rates etc, but it does seem that my old dinosaur home computers last longer than the newer PC's we have at work. I'd be curious if "power saving" is putting our data at risk.

Saving Power Has a Cost

[Prototerm]

The aggressive power saving settings here are perhaps a little too aggressive, but did anyone really think you could do that totally without cost? This isn't magic, you know. It's a trade-off. If you tell your computer (usually in a laptop) to spin down the hard drives to save power, you're going to cause greater wear-and-tear on the things because each time they spin down, they have to spin back up before you can use them again. If you want to save energy without the wear, turn the bloody thing off when you're not using it.

What, you're in too much of a hurry to view the latest pr0n? Chill, dude, before you go blind!

If you run a desktop, it's doubtful you'll have a problem with this, as most desktop users turn power saving features off entirely (and yeah, I also drive a big honkin' SUV. Bite me), but be careful on a laptop. If your hard drive supports SMART, you can do a quick check of the numbers (I think the one you want is # 193, IIRC), and see if you're at risk. But not all drives support SMART (I have a laptop drive that doesn't), so as usual, YMMV.

So Ubuntu can ruin hardware?

[Boojumbunn]

You know, it's funny. A while back there was an outcry about manufacturers who decided that installing linux nullified your warrenty. This incident makes me wonder if maybe they have a point? After all... they have likely tested the hardware for long term windows reliability. They probably haven't tested their hardware for long term Linux reliability (through all the various linux types and settings.)

That said, they could probably still support their warrenty on things they know won't be affected by operating systems, like the hinge of the laptops screen.

Boojum the brown bunny

Re:The Ubuntu

[Kjella]

I actually disagree. Sure, if the manufacturer issues a recommendation or something to change it, then by all means. Otherwise Ubuntu should just keep their hands off and let the HDD manufacturer deal with it. Does Windows automagically disable this behavior?

Same thing on Slackware 11

[Maestro485]

I have Slackware 11 on my laptop and I just checked the Load_Cycle_Count with:

smartctl -d ata -a /dev/hda

Currently the count is up to 1195740! So either I have the most durable drive ever created or this thing is going to explode soon. Does anybody have any suggestions on this? I don't know much about acpi.

Re:The Ubuntu

[cmowire]

I tend to think that somebody ought to take a fine toothed comb through a fairly standard desktop linux setup with a fairly standard set of applications and daemons and ruthlessly examine every disk access.

It seems like most of the things that are desirable for a server that are merely OK on a desktop are probably really bad for laptops and there's optimizations to be made.

And I'm not just thinking of Linux here. When I use a Windows laptop, I notice the hard disk spinning up on a fairly regular basis, even when I'm doing something fairly lame like web browsing or word processing. And you know that pretty much nobody thinks about it without being prodded....

But with Linux, you could make "Lapbuntu" that would contain a set of apps that were modified to aggressively avoid using the disk unless it's already spun up by patching existing software.

Re:Selected Excerpts

[skintigh2]

Not to defend any sloppy programming or unnecessary ware on hardware, but is this really that big of an issue?

At that rate, it will take you 8571.4 hours to hit that limit. For an evening user like me, even at 4 hours a day every day that's 2142.9 days, or 5.87 years. I'm used to Maxtors dying after 3 or so years, and my Seagates are usually obsolete (or dropped) in that amount of time, and that's on a desktop. I'm assuming not many laptops survive 6 years at all, or at least are used regularly that long.

Even a road warrior using it 60 hours a week would take 2.6 years, and it hasn't been my experience that laptops survive long with that kind of use.

PS: this article really should have been called "Ubuntu considered harmful"

Re:Thanks for slashdotting launchpad, guys.

[darkwhite]

The spindowns are not the problem! -B 1 causes many drives to wait a very short time before unloading the heads off the platter and onto the ramp. But the OS sends read/write requests to the hdd every few seconds, unless laptop mode is enabled. The drive then must load the heads again, service the request, unload them, etc. every few seconds. This is not as bad as spin-up cycles, but very bad nonetheless.

Ubuntu FOUND the problem

[darkonc]

One of the comments notes that at least one drive had the same problem with Windows -- The difference is that, lacking ubiquitous SMART tools, Windows users would be much less likely to get to the source of the problem.

So, it comes down to: Ubuntu users were able to diagnose the problem, and have the tools to implement a workaround. Nix to either for Windows users -- they just need to remember to replace their drive once a year.

Re:Old news???

[cuby]

it's even worse than that: http://paul.luon.net/journal/hacking/BrokenHDDs.html [luon.net] The article is from Nov. 25, 2005!

Re:My experience

[Anonymous Coward]

I did that test. Ubuntu 3 or 4 load/unload cycles per minute. Tried every hdparm setting I could think of without any visible effect.

Windows (2000 professional), normally around one load/unload cycle every 3 minutes or so. Sometimes it seems to go to sleep completly and does not load/unload for a long time until I resume use.

The Windows behavior is more or less reasonable and will extend my HDD life at least for a few more years. Ubuntu's behavior is a killer and I cannot tolerate it.

I have the feeling, but cannot confirm, that the problem with ubuntu is not excesive parking per se but that it unparks the head almost imediatly after parking like if something in the OS was accesing the HDD inmediatly after parking (my HDD was mounted -noatime so atime was not the culprit)

Re:no problem, really!

[Trenchbroom]

This is EXACTLY what happened to me. I had two Hitachi hard drives fail in my one-year-old Dell E1505. One came brand new with the computer, and one as the refurbished replacement under warranty. Both made a very steady "clicking" sound every 30-60 seconds right up until they failed. Also when I powered down the system the hard drive would make a very loud click sound upon shutoff. I read a thread on Ubuntu's forums that the clicking sound at shutoff was due to a flaw in 6.10 and 7.04 and that 7.10 would fix the problem. So when I received my Fujitsu refurb drive from Dell I decided not to install Ubuntu until 7.10 came out. Right after 7.10 came out the newsgroups started referring to the problem listed above. I ran the check to see what my cycle numbers were on this new hard drive--less then 3000 cycles with a usage of over two months. I completely believe that the Hitachi drives are too agressive with their settings and that the Fujitsu drive does not have the same problem.

Re:The Ubuntu

[cygonik]

Yes, it is a default set by the manufacturer. The problem is that Ubuntu touches the hard drive on a regular basis, causing the just-parked head to unpark.