Just wondering, how long time have YOU gone without rebooting your home/desktop PC? One guy I know has more than 450 days and counting on his, running CentOS.

I started my quest for something similar, now with a staggering 8+ days and counting.

How about YOU?


Best off-topic regards,
Strid

i don't generally pay attention to it.

i leave the computer up, and reboot it when a patch demands it.

i've system uptimes in the neighborhood of 2-3 months before, but his power cycle, the computer's only been on for 7 days.

this is running windows 7.

A few months at the most, just as poster above writes I too reboot when an update needs it. Win 7 is the OS.

450 days sounds like a very insecure centos system

I reboot my workstation whenever I need to (usually after an OS update).

three weeks at most .... it's quite difficult to keep my computers up in long period since -annoying- power outtage occurs frequently in my country

Aware of a remotely exploitable kernel bug? Because that's the only thing you need to reboot for.

Around two months. Usually, I just restart for Windows updates. The file server (unRAID), however, hasn't been turned off since April last year.

18 hours....when I forgot to turn it off the night before.

there's some program that allows linux users to install and use updates without rebooting the OS. i forget its name, but it saves you the downtime of rebooting.

Well, he only got the darn thing those 450 days ago...

You get the idea..



It's called a package manager and a sane design. ie. Not Windows. Windows is pretty much unique in requiring a reboot for every single update.

no, not a package manager. i know what a package manager is.

this isnt released under GPL. it's paid-for.

it's used for servers so that they minimize their downtime.


here we are.. http://www.ksplice.com/

Maybe so, but when a kernel update comes through on my Ubuntu system every few weeks, I'd rather spend 5 minutes mindlessly rebooting it, rather than 30 minutes poring through kernel changelogs to determine if I need to spend 5 minutes rebooting.

There are security announcements y'know.

I wouldn't trust that as far as I could throw it. Especially as they seem to think Dreamhost are experts. A properly configured server will reboot in a matter of minutes, for free. Less probably, as you don't have to reboot the hardware any more.

that is not correct, windows does not require a reboot for every update.

I gave up that stuf fon the 90s..
unix kernels of course were impressiveon simpl machines.

today it is better, but no long runtimes, I choose no to.

windows 7 even logs a "boot optimization defrag"..
at least it is honest.

Windows uptime always gets ruined by updates. Longest I've gone is the longest interval there has been between critical updates, which incidentally isn't long. 2 months?

Ok, more or less every update. Meanwhile I don't have to reboot more than once every four or five months.

Required reboots are not really a problem anymore, on Linux or on Windows. And who gives a shit anyway? This is a non-issue if you ask me.

I usually let my workstation (Win7) go to sleep on its own after a while, but sometimes even the modest 3W bothers me so I turn it of completely quite often. I seldom reboot just to reboot. My server (Linux) never sleeps and after a kernel upgrade I usually reboot it.

They're a problem for me for a number of reasons. One is I lose my sample rate without the SPDIF source, which means no audio at all while one box is down. Another is that it's unnecessary wear on mechanical parts (fans, HDDs, etc).



Well it's not an upgrade without a reboot.

i apologize in advance for sounding like i'm ragging on you but that is a pointless goal... there is absolutely nothing wrong with rebooting a server.



rebooting a server doesn't mean that you have to shut the power off, but...

how many times have i argued this with the engineers that i used to support... if the voltage spike on cold startup was an issue, how come the manufacturer(s) never have a specification for power cycling?

if it was a significant failure mode, there would be a limitation on it.

I have never seen an incandescent light bulb (or even florescent) burn out while it was on (although it can happen), but have frequently seen lights burn out when first turned on. So there definitely is some issue with turn-on spikes in electrical equipment.

But I am not sure if it applies to DC to the same degree as to AC. Also don't know if having a UPS and good PS can effectively prevent turn-on spike problems. However, even if it can happen (turn-on spikes damaging equipment), doesn't mean that it is anything to worry about in most circumstances for most parts. I do think that disk drives are an exception and there is some concern for them, which is why disk drives are rated for number of turn on cycles.

It's not a goal, it's a reality. I do not need to reboot, because there are no updates which require it.

then you shouldn't have any problem showing us that rating, for the WD2001FASS that i just bought today.

it's a totally irrelevant reality; about as important as how cold the moon is.

In checking the WD website, it appears they no longer publish on-off cycle ratings (at least that I can tell on their website). They do publish rated number of load/unload cycles (300,000), which is somewhat related to on-off cycles (both of which are tracked by S.M.A.R.T. the Self-Monitoring, Analysis, and Reporting Technology).

Seagate also rates many of its consumer drives for 300,000 load/unload cycles, and rates them for 100,000 on-off cycles (but the on-off cycle rating depends on humidity and temperature).

As to why WD no longer rates their drives for on-off cycles (even though Seagate does), I don't know, but probably because it not a "practical" limitation. If a drive is rated for 100,000 on-off cycles (as Seagate does), then that would be 27 reboots every single day for 10 years, something that is extremely unlikely to happen. Also, the 300,000 rating for load/unload cycles is likely to be reached before a 100,000 on-off cycle limit is reached (since a reboot includes two load/unload cycles).

Although I stated my opinion in a previous post that rebooting a modern computer should not be a concern, it would be absurd to suppose that one can reboot a computer an infinite number of times without some adverse affect to a hard drive (even if the drive did not completely fail).

If the computer is rebooted, not shut down and then turned on again, then the HDDs continue to spin throughout the reboot cycle. Unless the HDDs are spun down to begin with; I've noticed that my disk server spins up all the HDDs one by one before a reboot.

you still don't get it, so lets try another approach.

here is the seagate warranty: http://www.seagate.com/ww/v/index.jsp?l ... 04090aRCRD

do you see anything about power cycling in there? of course not.

a drive can be set to spin down during sleep mode, in order to save power, so the heads will park at that point, it's not just when the power is turned off.

I said that drives are rated for start-stop cycles. It turns out that WD no longer publishes that spec, but Seagate does. I also mentioned that the load/unload spec that WD publishes will likely occur first, so that may be why they don't publish a start-stop cycle spec (and also because it would be extremely difficult to actually test it).

I don't understand what the warranty has to do with this discussion (you will have to explain that to me).

I said previously that start-stop cycles is not a practical limitation, since no one can turn off and turn on (cold boot) their computer 100,000 times (the Seagate rating for their consumer drives). But the Seagate drives are nonetheless rated for start-stop cycles.