The Ubuntu operating system has been charged with crimes against hard drives. A number of users have complained this week about the OS (7.04/7.10) forcing drives to spin up and down at an unnatural rate due to some very aggressive power management features. According to Ubuntu wizards, however, this is a firmware/BIOS issue and …
As OS that won't screw with the hardware?
it just uses it the way the manufacturer intended?
gee... what a non-microsoft concept.
Bad drive firmware implementation all around
-B and the APM register setting should control the overall drive APM behaviour not spindown times. The spindown is controlled by a different setting -- -s. So if a drive spins down straight away on -B1 ignoring whatever it was told using -s it is the drive manufacturer's fault, not the OS fault.
While at it, if you want to control your drive power consumption besides -s the other important setting is acoustic management (-M). The defaults for most drives are "max speed, min power saving, max noise". Turning this one (along with a sensible spindown time) makes much more difference that touching the APM register.
so there is a good reason to keep reading El Reg apart from Paris Hilton news!
I've been fooking about with this latest Ubuntu release on my cr*ppy Dell cr*ptop for days, trying to get it to work - now I know why it doesn't work.
Now I can go back to reading Paris Hilton news.
Maybe not just Ubuntu
Since I installed 1 of the performance & compatibilty patches on my Wndows Vista a similar thing has been happening with my second hard drive (spinning up and down) I will now have a play with my bios settings and see if it goes away.
Linux is like a fresh from University computer scientist all full of best practice, open standards and protocol theory. Windows is the hardened admin who has been around the block and knows that in the real world you don't always just leave things to happen the way the hardware vendors intended.
I loved the little excuse about it being a bios problem which everyone else fixes but Ubuntu doesn't so clearly it's still a bios problem rather than being something which Ubuntu could have fixed as well.
That said, it's also incredibly naive to leave the settings which determine the life of the hardware in the hands of the people who make the hardware... a cynic would suggest that they *want* it to fail early.
This is rumor control...
...here are the facts.
First, it is not accurate to say that "When switching to battery power, /etc/acpi/power.sh issues the command hdparm -B 1 to all block devices". This is not the default configuration of Ubuntu. It only happens when laptop mode is enabled, which must be done manually by the user.
In general, Ubuntu does nothing to hard drive power management settings. It leaves them exactly as it receives them, from the machine's BIOS. If your machine's BIOS sets a ridiculously aggressive power management strategy, that is the fault of its manufacturer.
This also applies to Mandriva and, I believe, to most other distros.
The suggestion that "Of course, your hard drive my eventually catch on fire, which would also lessen its lifespan." is also silly. Take a look at the specs of a typical laptop hard disk:
*Maximum* power consumption, at power-on time, is a princely 5W. Consumption in active read / write use is 1.8W. Consumption in active idle mode is 0.8W.
An absolutely maximum consumption of 5W and a more usual consumption below 2W is not going to set anything on fire.
Windows Vista is doing this too....
Another blogger has also run tests comparing load/unload cycles of Windows Vista and Ubuntu.
"Now I rebooted in Windows Vista Home Premium, let the system run for fifteen minutes and rebooted into Ubuntu. After taking measurements again, surprisingly there was a ten cycle increase.
So the only conclusion I can draw here is that Windows Vista does not alter settings provided from BIOS/microcode either."
Re: This is rumor control...
Er, the fire thing was meant to be silly. It's a joke. Remember all those exploding laptops?
It's not the spin down that's the problem. It's parking the heads every few seconds, then re-loading them to the disk, a problem particularly with those drives that have an "off ramp" for the heads to park on I'd imagine. The SMART attribute for this action is 193 Load_Cycle_Count.
And no, this isn't an Ubuntu problem. FreeBSD on a suitably equipped laptop leaves the APM alone as well, leading to attribute 193 incrementing every few seconds and the tap-whir-tap-whir noise from the hard disk. sysutils/ataidle is the cure here, using the -P 254 flag. It's similar to the solution proposed by Linux-Hero with hdparm -B, although 255 doesn't work for some hardware and you're better off using 254.
Nor is it new. Here's a link to some poor bastard in 2004 with five hundred thousand head load cycles on his hard disk. Not surprisingly, it has failed SMART.
Catching on fire...
...probably comes from the battery starting the flame front. Yes, something as small as 5 watts CAN start a fire (there are things smaller). The controlling thing here is the dissipation of the heat product. Luckly a disk drive has a pretty reasonable heat sink (most are metal enclosures), so there is little chance of starting a fire. The part that catches fire is most likely a circuit board, but most of them are flame proof (FR-4 material). I suspect that the result of a fire on a disk drive would be some scorching, but I have yet to try it. Maybe one of these days I'll throw one on the barbie :-).
Surely one of the open saucers can provide a "control panel" to set this the way that seems best.
btw XP seems to access my hard drive every few seconds even when doing NOTHING!
WTF with that Micro$haft?
Blame anyone you want, but if it smells bad only when you are in the room.
> Linux is like a fresh from University computer scientist (...) Windows is the hardened admin (...)
So presumably, in displaying this astonishing ignorance - you'll just be the apprentice floor cleaner speaking out of turn.
...with linux zealots making mac zealots look calm, I'm taking my life in my hands here, but jeez - talk about cutting off your nose to spite your face. Again - Ubuntu is supposed to be "linux for real people" - as their live CD jacket says - but then you're supposed to run a cryptic command prompt line to get it to tell your hard drive not to die? All this because the holier-than-thou creators of Ubuntu are sputtering - but but, it's not OUR fault - despite their making a conscious decision to cause the problem.
Sounds like 'linux for linux nuts' to me.
When I run an operating system, I run it to accomplish things. Windows is far, far, far from perfect - but I don't have to put up with windows deliberately failing to patch known hardware problems to make a theoretical point. And if you ARE going to make a point, let it get on all the IT sites and then release an easy to use patch for the "real people" using your distribution to do something aside from being self-righteously upset.
The whole thing strikes me as being rather like a group of people who, believing that handguns should have better safeties, give a boxful of revolvers to a kindergarten class and blame the ensuing bloodshed on the manufacturers. Yeah, maybe things *should* be different, but most of us aren't arrogant enough to invite disaster merely for the pleasure of being right.
There is a clear fix here...
Just upgrade to SSDs. It isn't the fault of the Ubuntu devs that you all still use ancient spinning media - pony up the $600 or so and get with the times.
This is really just another of those Ubuntu 'issues' that are really only problems because their users don't know much about how to work Linux. Things happen automagically and the user doesn't get involved. On Slackware when I wanted more battery life I read the hdparm man page and fixed everything up myself.
Linux is like a fresh from University computer scientist (...) Windows is the hardened admin (...), indeed. I think the IRC community said it best:"ROFLMAO".
The day Windows ousts Linux as the "hardened admin", I will eat my laptop. Windows is only good fro playing games, aside from that, it's all window-dressing (excuse the pun).
Using the command prompt
..is all very well, but as David Wiernicki states, Ubuntu is supposedly created for the masses, and not everyone is confident tinkering in terminal.
That said, I am pretty sure that if this really is an issue then a fix will be posted to the repositories in the next few days.
This was written using Ubuntu 7.10 on a PC largely composed of laptop components and no ticking hard disk.
As has been said earlier, your beloved Vista also suffers the same problems. The only problem there is you have no such control over turning it off, whereas with Linux you do. :-P
I groaned when I read that, and I'm a Windows fan. I can understand where he's coming from (in regarding the metaphor in context), but when the context changes even a little it becomes a ridiculous statement.
Also, to paraphrase Nexox Enigma:
Just upgrade. It isn't the fault of the Vista devs that you all still use ancient PCs - pony up the £500 or so and get with the times.
As Linux fanbois and Jobsies alike are constantly telling us, you shouldn't have to upgrade a PC, EVER, just to run an OS.
Vista does unusual stuff to the HD too.
I've just installed Vista on a machine here at work. I've had it at home for ages, but I never use it when I've got Ubuntu for normal use and XP for games.
Anyway on this work machine it takes a full 10 minutes after booting up for the HD to stop chuntering away. I've been monitoring and Vista is just constantly going through loads of random files, checking dlls, fonts (for some reason), writing loads of stuff to temp, and to 'softwaredistribution', and 'wbem\repository'.
I know this isn't the same as unnecessarily spinning down the drive and parking the heads, but it's annoying nonetheless.
Now proceeding to the cloakroom as per applicable rules.
well done JP Strause
I would also like to add;
what a fucking cock!
It is sort of Linux fault
First, I fully support Linux. However, if Linux developer knows certain BIOS/firmware setting harms hard drive, it should control the harmful action by override BIOS/firmware.
It is technically not Linux fault, but who do you think the uneducated users will blame?
On the other hand, disable this crazy "power management", the same uneducated users will than blame Linux for "using more power".
You cannot win, but I'd rather choose the less blame.
I see a lot of opinions here.... but as I type this on a Tosh laptop running Ubuntu 7.10 I don't hear my HD in its death throes. Neither am I a "hardened linux nut" I have an MCSE (work demands it and paid so who am I to argue) but some tools I want to use work better, or are open source on linux so I thought it was about time I took the plunge and started using Linux as my main OS on one of my machines...
IMHO if this is only a problem on a couple of manufacturers machines then how can it be an Ubuntu problem or even a Linux problem...
People just cant help themselves but pick a side and fight over it... now which end of my boiled egg shall I cut off today?
it wasn't me
Lets face it there are already millions if vista laptops out there and have been for some months now and no problems with reagards to the power management of the hard disks being reported, unbunto has a install base on laptops of what? lets say 10,000 (no where near that much i bet ) of sucessful installations and all these reports comming in. from the above facts i think we can come to some conclusions
1 Vista isn't doing it as well
2 unbuto has a serious bug which may cause hardware failure which needs sortin. asap
vista did it too sir! Please thats just typical of linux fanboys who's staple diet is Pizza and own a clingon suit, what's the klingon for lonley? get a life or actually better still get a girl friend!
This seems to me to be a clear case of the Law of Unintended Consequences.
Modern laptops (thanks largely to ever-increasing hardware demands from Redmond) have plenty of RAM. They also have hard disks which go into a power-saving mode (with the heads parked and the motor off) after spending some predetermined (and theoretically adjustable, but in practice almost nobody bothers to) amount of time idle.
Linux by default uses a highly-aggressive disk-caching policy: basically, never, ever write anything to disk unless (1) the machine is about to be turned off or (2) there is no RAM left to cache it. Entire temporary files can be -- and often are -- created, read and deleted without ever going near oxide. (That, incidentally, is why some software runs blisteringly fast on Linux, but crawls on other OSes with different caching policies.) Aggressive caching + much RAM = fewer writes to disk.
Now, when you combine a hard disk drive that automatically goes into power-saving mode with an operating system pretty much designed to leave disks standing idle, what should you expect to get?
My guess is that the reason why this only seems to affect Linux is that Windows feels the need to access the disk just often enough for the drive never to go into power-save.
One final thing: When starting from rest, a HDD motor draws several amperes for a brief instant, as it has to accelerate the disc platter suddenly from stationary to several thousand RPM -- but once it's running, the current drops dramatically as it only needs to overcome friction in the spindle bearings. It's entirely possible, by injudicious settings, to end up wasting more power on standing starts than you saved by stopping in the first place.
why not go for flash hdd instead??
why not swapping all moving parts from our computers for active coolers??
why not getting rid of this unreliable mechanical oh-so 19th technology all together??
hardware makers, please catch up...
My clingon doll gives me inflatable lovin'
I guess this shows the open source difference. A problem is reported with Ubuntu and within minutes is being discussed worldwide. The technical root cause is more-or-less pinpointed, workarounds provided and strategies proposed. The code is scrutinized and its authors questioned.
Were the same thing to happen in the closed source world, the problem would be promptly denied by a PR machine. Executives would be wheeled out to obfusticate, and months (or years) later, a fix would be quietly smuggled out.
So is Vista bothering disks like Ubuntu ? Well, since the code is top secret, we don't know. And while all PC buyers are forced to also buy Windows, there's no chance of it harming sales, so don't expect Microsoft to act quickly.
Incidentally did Microsoft ever get round to fixing that Vista file copying/deleting bug ?
I must not remember where I wrote that I loved Vista. Perhaps it was "Windows is far, far, far from perfect"? Anyway, I don't run Vista; I run XP.
I run XP because only XP and OSX have the software I need to run without requiring massive amounts of knowledge to set up, and OSX won't run the game software my business is based on (not to mention the hardware tax).
So, yes, I run XP. I don't belove it, and I don't run Vista at all as I don't trust it. But Windows does let me do my job, and doesn't have a group of cheerleaders telling me to go screw myself at every turn because I'm not geeky/cool/smart enough. Hell, at least Steve Balmer is good entertainment.
Compliments to Ashlee and everyone
I feel a bout of nostalgia for the El Reg of old coming on.
@A. Lewis: how is your anti-virus scanning set to run? At start-up?
I see a MS vs Linux flame war erupt in Reg comments I notice that the Linux fans attack the Microsoft fans' Operating System's security, stability, efficiency, privacy, cost, ethics and so-on while the Microsoft fans attack the Linux fans' sexuality, fashion sense, body odour, social skills and hairstyles.
I think this probably says all you need to know about Microsoft, Linux and their respective fans.
as anyone bothered
to check if the drive also performs a spindown ?
head load/unload and spindown ae separate events...
the firmware in a harddisk does not unload the heads by itself. this has nothing to do with drive firmware. This control command is issued from the PC. whether it be the BIOS or the OS.
I seriously doubt its the bios ... after all once the OS is loaded the bios is pretty much dormant. i doubt that ubuntu / linux / windows are still performing bios calls... it may be a setting in the configuration that is beeing read and used , and that is a whole different story. that is a matter of reading the settings , throwing out what is ridiculous and using them with half a brain... just because some dimwit ticked the option to spin down every 5 seconds doesn't mean the OS should do this. It could display a warning box : are you sure you want ot put it that low ? this causes wear and tear .... blabla
i guess the fanboi's are too busy debating whether to use Gnome or KDE and Emacs or VI (lunix : deliberately misspelled ) ... or to ram aero and mandatory 'allow-deny' and secret updates to the updater down the users throats (windoze) or force you to use At&T for fear of bricking your iAddiction ... (didn't wan't to leave those out.)
me ? i'm waiting for the paris-hilton os ... so i can upgrade my DOS3.2 box ...
Not a Ubuntu problem
It looks like this really isn't a Ubuntu problem - apparently, it happens on Windows too. The reason people thought it was associated with Ubuntu is that it comes with tools to query the hard disk's head load/unload count, and Windows doesn't.
BTW, the last time "after all once the OS is loaded the bios is pretty much dormant" could ever actually be true was the 286 era, back when the BIOS was still used for everything anyway. (Look up "System Management Interrupt" or SMI sometime.)
Also, it's not the OS spinning the hard drive down every 5 seconds. Remember, modern hard disks are essentially independent devices with their own processors - it'll be the disk spinning itself down, without any OS intervention. (All modern hard disks have an option to do this.)
That is probably the most insightful comment anybody has ever posted on any forum anywhere ever. Well done!
common admit it it's friday! its a major bummer for unbuntpoo i'm sure it'll get sorted out, before you can fix a problem you must admit you have one :D
regarding the attacking of Linux fan boys with regard to their sexuality i would never do that neiter would i accuse them publicly of being mummies boys who have never kissed a girl no in fact i would see that as imature and generally un called for. as for all the klingon references it is everyones right in a free society to be a sad git appoligies if i implied that wasn't the case ealier.
p.s my only pet hate with linux is that it never works quite right, i'm getting me coat, quick!
Sorry, but it IS an O.S. problem.
The goal of an OS is to manage the hardware. If the hardware is doing stupid things, the OS needs to get in there and override the settings if it can.
So yes, it's an OS problem. The Linux crowd is going to have to learn to get away from putting the onus on users to fix "well known" issues with hardware. Especially since they can do it so easily on boot because the OS provides the tools to do so.
On a different note...
...it might nice if, for once, the Microsoft crowd would desist from using "fanboi" when referring to Linux or Mac users, with obvious reference to Japanese "yaoi" or boy-boy hentai manga. Yes, the Linux/Mac proponents are rather over-the-top rabid in their opinions. However, using the term just makes the Microsoft crowd look like Bill Gates wanna-be towel boys picking up soap in the shower. Now, the just as rabid Microsoft crowd might enjoy being raped by Gates & Co which is why they keep on bringing up the gay sex theme, but it really doesn't do their cause much good to continue reminding everyone.
ashlee re humor
ah, sorry about that, ashlee. you forgot your SARCASM tags. =) but seriously - it really didn't read like humor at all...to me anyway.
The funny thing is that absolutely no-one has recorded an instance of a hard disk dying due to this problem. The whole kerfuffle was just kicked off by someone looking at smartctl output. No-one bothered to check whether this is anything that's changed recently, what it's due to, whether it happens on many systems, whether it also happens on other distributions, or whether it also happens on Windows. So it was basically one big giant ball of confusion. Now people are starting to do those checks, in a haphazard way, which is contributing to the confusion as different people have different levels of knowledge about the whole issue.
so, basically, unless you bother to go and read the useful posts by the people who actually have a clue what the hell they're talking about - the Ubuntu maintainers and other people actually involved with this stuff directly - or, even better, go and test it for yourself - it's really a bad idea to comment on it.
Does this happen with laptop-mode disbled?
I am using Ubuntu 7.10 on an ASUS A6Rp laptop. For a short-while I *manually* enabled laptop-mode. I found that it did indeed park the hard drive and power it down. I found that the time to park and unpark, seek and read info then start whatever application I was trying to use was annoying, so I turned off laptop-mode.
My point is that I had to *manually* enable laptop-mode, and *manually* disable it, to recreate the behaviour that is being discussed. It does not happen otherwise.
Are the people who are having this issue getting it with laptop-mode disabled?
Re: Every time...
Perfect, perfect... Should be nominated best comment of the year.
> I run XP because only XP and OSX have the software I need to run without requiring massive amounts of knowledge to set up
Same old FUD as usual here and your earlier post, David. "massive amounts of knowledge to set up" hasn't been true for Linux for ages but of course if you could only extract yourself from your Windows handcuffs and stop acting the shill you'd know this. I seriously doubt you've *ever* tried to install Windows onto bare metal.
what the watt?!
Adam Williamson: 5W is huge! my laptop uses 11W on battery, I will definitely notice the difference of a few watts.
don't believe the raw value
when you run smartctl on your disk and look at the Load_Cycle_Count entry, and the RAW_VALUE column, you may very well *not* be seeing the number of times your hard disk heads have unparked.
SMART is not designed to reliably report counters, instead it reports an indication of health.
You will see VALUE, WORST and THRESH columns - these are the useful ones. Load_Cycle_Count is an Old_Age type of SMART value, which means the VALUE starts at 100 and counts down to THRESH (usually zero).
When that counter reaches zero, it means you have reached what the manufacturer of the drive believes is the end of its expected lifespan.
By way of an example, my thinkpad x40 reports a Load_Cycle_Count RAW_VALUE of almost 3 trillion, which for a 14 month old laptop is very clearly not a real counter. The VALUE is actually only about 71, so I've used less than 1/3 of the drive's life, even though I've had aggressive laptop mode enabled in Ubuntu for at least a year.
Some drives will report the absolute count of head unparks in the RAW_VALUE column, but please don't assume that it is always the case. Use SMART for what it is supposed to be, an indication of health that is interpreted by the drive's firmware (ie by the manufacturers). That is why there are thresholds and stuff.
Sadly this entire issue has blown up because of a mis-interpretation of data and some broken assumptions about what Ubuntu does when you are on battery.
- Analysis Who is the mystery sixth member of LulzSec?
- Analysis Hey, Teflon Ballmer. Look, isn't it time? You know, time to quit?
- Murdoch Facebook gloat: You're like my $580m, 'CRAPPY' MySpace
- Tablet? Laptop? HP does the splits with Tegra-based SlateBook x2
- NASA signs off on sampling mission to Earth-threatening asteroid