View Full Version : Can I run it 24/7?
cypher543
2006-09-04, 07:33 PM CDT
I read somewhere that a computer should not be run 24/7. But, I never really knew if this was true. So, my question is simple...
Does it hurt anything to leave my computer running 24/7? Will it burn something up or fry my HDD?
bob
2006-09-04, 07:44 PM CDT
Many people do leave the machine on at all times without problems. Of course, there's more power consumed and therefore higher electric bills.
commonplace
2006-09-04, 07:47 PM CDT
There are two different schools of thought. One way of thinking is that the hardest thing on a computer is the startup, when it jerks to a sudden start, which wears down the hard drive, etc. So, by that way of thinking, leaving it on 24/7 is best.
The other way of thinking is that leaving it on 24/7 is constant wear and tear, especially all that time when you're not even using the computer.
It's been a few years and I can't cite the exact article, but it was PC Magazine or something similar that did a case study on it (or at least reported on the results), and basically it was a wash -- neither was significantly better or worse than the other.
For me, it comes down to a matter of convenience. I leave mine on 24/7 because I like to be able to sit down and check e-mail, surf the web, etc. without having to wait for the computer to boot. If you use your computer frequently, that might be the best route. If you only use your computer once a day to check e-mail for a few minutes, then leaving it off when not in use might be best for you.
I'd say it's just up to each individual to decide what's best for themselves.
Just remember, servers do nothing BUT run 24/7. Google, Microsoft, and yes, Fedoraforum.org, all run on servers that are turned on and receiving heavy usage 24/7/365. Would they last longer if they weren't left on all that time? Dunno. :)
pparks1
2006-09-04, 08:04 PM CDT
I leave 4 computers on 24x7 in my house. I work with servers and they run 24x7.
I'm of the mindset that constantly warming up the components and cooling them off is not good for them. I would rather get them up to standard operating temperature and keep them there. So, that is what I do.
cypher543
2006-09-04, 08:18 PM CDT
Thanks for the info, guys! All I need now is a computer that doesn't sound like a train. :p
Swish
2006-09-04, 08:19 PM CDT
There are always caveats when answering this type of question.
Things will not generally "burn up or fry", if they run at an acceptable temperature.
Make sure all your fans are working properly and not making any 'unusual' noises, and that the machine reasonably clean inside. Proper spacing of components to allow proper heat dissipation maybe warranted. If you are still worried, add an additional case fan.
I converted an old desktop (AMD Duron 1GHz) to a 24/7 home server a few years ago. I changed the PSU since the fan was noisy, and set the temperature warning settings in the BIOS. It runs fine, and is only rebooted for kernel upgrades. Once a year I clean it out with a can of compressed air.
cypher543
2006-09-04, 08:28 PM CDT
Once a year I clean it out with a can of compressed air.
Is it safe to just open the case and blast the whole thing with the compressed air? (while it's turned off, of course) What I mean is, would it be better to remove each component and clean it seperately?
Swish
2006-09-04, 08:48 PM CDT
Taking each component out sounds like too much work :)
Seriously, the compressed air can reach where the normal air flow goes, and my aim is to remove any dust build up. If I remove components I risk blowing debris into slots that I will push cards back into. Handling the components also increases the risk of static or physical (oops :eek: ) damage.
If you are comfortable taking it apart for cleaning go for it, I have however heard of older systems being pulled apart and not working once they were put back together.
Qwagmire
2006-09-04, 09:15 PM CDT
I've heard, though anecdotal, that the cpu has the best longevity at a relatively constant temp (continually on). The analogy made is that it is like a light bulb; a light bulb will last longer if continually left running current as opposed to the dramatic temperature changes with turning it on and off, which eventually breaks the filament.
In the analogy, continually starting a cpu from a cold temp could hasten changes to it.
It's compelling, though I'm not sure if it is true. Perhaps someone here will shed some light on it.
jasonscsn
2006-09-04, 09:30 PM CDT
I always leave mine on 24/7 for the fact that someone said. I use the computer many hours a day off and on, not constant. Therefore Its pointless to turn it off after each time because 5 mins later I might be like "Oh wait I wanna check up on this" Then Have to wait for it to boot etc.
I even told my grandparents who always used there computer(only to check email and browse of course) to just leave it on after they were done cause it makes things much easier when you want to use it again. Well eventually they started doing that.
Dan
2006-09-04, 09:59 PM CDT
Evenin' cypher.
You're getting a lot of good info here. Here's my two cents.
If running Windows, I shut down at night. (and under hard use, reboot at least twice daily.) Running Linux, it doesn't bother me to leave it up 24/7/365. I only reboot RH9 if a program crashes hard or I get a fragmentation / segmentation fault error. (It has a lot to do with the way each OS handles memory and disk usage.) In either system, an illegal operation or a segmentation fault is good cause to reboot. A computer's memory can't be just slightly corrupt any more than a stray female cat can be "just a little bit" pregnant. It either is, or it ain't. A reboot is cheap insurance. :cool:
One more caveat. The fans in your system act a lot like a vacuum cleaner. In comes the dirty air, out goes the filtered air! The remains (dirt, dust, cat hair, volitized smoke components, owls, tree branches etc...) coat the components inside the box. The more that junk builds up, the hotter everything runs. The hotter everything runs, the faster it dies. Canned air helps, but a thorough cleanout is needed regularly. (I take my box out to the garage and let 120psi air from the tire compressor do the work for me.)
However, know that too much dust stirred up in the box can eventually settle on the lenses in the CD / DVD optical drives. Then they will get cranky, and eventually fail to read or write. (BTW, smoking consistantly around your computer / optical drives is a sure way to improve the mood of your local computer parts dealer.) Use care blowing them out.
I do not recommend disassembly to clean. It ain't broke, it's just dirty. Murphy's law is still in charge, and during teardown and rebuild, something is bound to get unseated, missaligned or scratched and / or busted. I learned the hard way that a suddenly unseated CPU cooler will make a grown man cry. :eek:
Hope it helps...
Dan
ccrvic
2006-09-05, 03:06 AM CDT
I only reboot RH9 if a program crashes hard or I get a fragmentation / segmentation fault error. (It has a lot to do with the way each OS handles memory and disk usage.) In either system, an illegal operation or a segmentation fault is good cause to reboot. A computer's memory can't be just slightly corrupt any more than a stray female cat can be "just a little bit" pregnant. It either is, or it ain't.
I think you've missed out on what a segfault is.
A segfault is when the OS[1] has noticed that an application is trying to step outside its walled-off area of memory. That means that something has gone horribly wrong with *the application*, but all other memory is untouched. The app is killed instantly, before it can damage anyone else's stuff.
A reboot is cheap insurance.
A reboot is completely unnecessary, just because a suerland app has segfaulted.
The fans in your system act a lot like a vacuum cleaner. In comes the dirty air, out goes the filtered air!
That depends on the placement of your filters & the direction of your fans.
Vic.
[1] Strictly speaking, it's normally the MMU that will "notice", but it's handled under OS control.
ccrvic
2006-09-05, 03:11 AM CDT
Is it safe to just open the case and blast the whole thing with the compressed air? (while it's turned off, of course) What I mean is, would it be better to remove each component and clean it seperately?
I wouldn't.
Dirt ingress iis only a problem in three situations :-
Cooling systems getting blocked
Optical devices getting covered
Dirt getting into connectors
Of these, the last is by far the most likely, and the one exacerbated by unseating components.
Vic.
cypher543
2006-09-05, 06:07 AM CDT
Wow. Tons of info! Thanks!
Does anyone have a Foxconn TS001-V-A300A Black/Silver Steel ATX Mid Tower Computer Case w/300W Power Supply? If so, how quiet is it and does it have good airflow? That's the case I want to buy for my next Fedora system.
pparks1
2006-09-05, 06:32 AM CDT
I would suggest the Antec Sonata II case. It has great airflow, is very quiet and is a great case to work with. This is what I use on my Linux box at home and with the Sempron proc and the amd stock cooler, you can hardly tell the thing is running. It also has an air filter in the front to keep the dust out. It has rubber grommets that the hard drive sits on to keep vibrations away from the case. And it has a genuine Antec Power Supply...and those are very well regarded power supplies.
http://www.newegg.com/Product/Product.asp?Item=N82E16811129155
cypher543
2006-09-05, 06:35 AM CDT
Thanks for the suggestion, but I find cases with front access doors annoying. I have one right now, and I hate opening it to pop in a CD.
pparks1
2006-09-05, 06:38 AM CDT
If it helps, the door easily pops off.
I don't often use my optical drive on my Linux box, so I don't care that the drive is hidden.
Dan
2006-09-05, 07:28 AM CDT
I think you've missed out on what a segfault is. Morning vic.
Nope! I didn't miss out on a durn thing! :) What I have noticed through experience though, is that there is often quite a bit of difference between the textbook scenario and life in my little corner of the real world. :cool:
I will wholeheartedly admit that in Linux, often a reboot is not strictly needed. But sometimes it can save you a lot of grief dealing with an underlying problem that will only get worse the longer you fiddle with it, or anything else that the offending application shouldn't have messed with... but did anyway.
And, doing so won't hurt a thing, and also gives the wetware a good chance to get up and do a little re-setting too.
As to the second point, That depends on the placement of your filters & the direction of your fans. OK. I agree. But... On the thumbs of your left ear, count the number of filters properly placed (if they even exist) in the average user's computer. :D (heatsink fins don't count!)
Dan
ccrvic
2006-09-05, 07:45 AM CDT
What I have noticed through experience though, is that there is often quite a bit of difference between the textbook scenario and life in my little corner of the real world.
That's something you'll want to look at, then.
A segfault on one uerland programme can't trample on another - that's the whole point of memory segmentation. If you're seeing segfaults prior to a more serious problem, the implication is that the initial segfault was a symptom of something more serious - usually a problem accessing RAM. That's often because of dirt in the DIMM socket, or just the DIMM coming loose.
I will wholeheartedly admit that in Linux, often a reboot is not strictly needed. But sometimes it can save you a lot of grief dealing with an underlying problem that will only get worse the longer you fiddle with it
Disagree.
Rebooting all the time has serious consequences when there are many people trying to use your machine. Additionally, it usually masks an underlying problem - which then persists, rather than getting fixed.
or anything else that the offending application shouldn't have messed with... but did anyway.
But that's the whole point of the Unix security model - if it shouldn't mess with something, you don't let it. SELinux adds more granularity to this, but the core concept is still the same - fence all your apps into their own area where they can't hurt anyone else.
That way, you only need look after your kernel.
And, doing so won't hurt a thing
Tell that to the matey that's just lost a multi-day session...
and also gives the wetware a good chance to get up and do a little re-setting too.
Drink more fluids, That'll have the same result...
As to the second point, OK. I agree. But... On the thumbs of your left ear, count the number of filters properly placed (if they even exist) in the average user's computer.
I wouldn't know about "the average user's computer". I try not to play that game. But it's not exactly rocket science to work out which way the fans are turning while you're building something...
Vic.
Dan
2006-09-05, 08:42 AM CDT
Tell that to the matey that's just lost a multi-day session... Not an issue here. I'm the only pebkac on the sysytem.
Drink more fluids, That'll have the same result...
...it's not exactly rocket science to work out which way the fans are turning while you're building something... Especially with the little arrows embossed on the fans! I think these are related. Strictly wetware issues. And thanks for the belly laugh! However, I still maintain that the "average" user, albeit unaware, uses their system's internal componentry for filtering. The concept of a filtering element on the outside of the intake fan(s) seems to have failed to become popular.
But that's the whole point of the Unix security model - if it shouldn't mess with something, you don't let it. SELinux adds more granularity to this, but the core concept is still the same - fence all your apps into their own area where they can't hurt anyone else. That way, you only need look after your kernel. Hmmm. SELinux and I are still glaring at each other from opposite sides of the virtual room. It's up and running on the FC5 drive, but I haven't dug into it, so I don't necessarily understand everything I know about it yet. However, I do note through experience, that (on RH9, FC2, FC4 anyway) a hard Firefox or OOo crash can still manage to smack X down too. (And, therefore, anything else that was running on X at the time.)
Dan
vBulletin® v3.8.1, Copyright ©2000-2009, Jelsoft Enterprises Ltd.