Microsoft is working on a significant new feature for Windows Vista, known as Restart Manager, which is designed to update parts of the operating system or applications without having to reboot the entire machine. “If a part of an application, or the operating system itself, needs to updated, the Installer will call the Restart Manager, which looks to see if it can clear that part of the system so that it can be updated. If it can do that, it does, and that happens without a reboot.” And here & here are some more shots of Vista build 5259.
Welcome to the 20th Century MS.
LOL. Next they’ll patent “the ability to install new features and make it usable without requireing a computer reboot.” Like they try to do with other apps that have been around for decades before they did it.
First user appears to be an administrator. Yeah yeah, screenshot is of an unreleased beta, but still… you’d think that this far along they’d have switched to their new system where a regular user account is used by default. Eesh.
when will Microsoft’s Windows technology catch up with the Unix, BSD, and GNU/Linux world, not to mention a number of other OSs with the said function?
Welcome to the 20th century in what way?
Linux -> upgrade of kernel requires a reboot. Update of operating system (usually major revision jumpes i.e fc 3 to 4) requires reboot.
FreeBSD -> requires reboot for kernel updates/ recompilations and for the makeworld procedure
OSX -> reboots for upgrades and also for some driver installations (had a printer driver insist on rebooting for me)
Granted windows might need to do it a bit more than the others notably because usually applications under unix don’t usually require a reboot (although if they demand a kernel module, kernel feature or some other bizzare thing it may be required), but they all do it.
This is great news.
“Upgrade of Kernel requires reboot”, while Windows needs it for, err, almost everytime something is installed or settings changed!
(note, slight exaggeration)
Which part is the exaggeration?
A lot of apps request the reboot but it is almost never needed.
I’ve even heard of printer drivers requesting a reboot. It’s pure BS.
What are you talking about !!
=> on windows you reboot on :
– driver upgrade
– software upgrade
– freeze of system
=> Fedora Core … really what is this “linux distro …”
maybe a core dump distro
Anyway !
In Unix you can update userspace apps without rebooting because you can unlink a path name from its inode even if the file is already open. In Windows, you’re forced to wait until all the apps close that file. This means you can’t update things like ie: explorer.exe or IE without rebooting. In unix you can, and when you restart the APPLICATION (not the whole system) the app will be running the new version not the old one.
Microsoft, welcome to 70’s technology.
Interestingly, I’ve been told you can rename a file while it’s in use, then swap the new file, and delete the old one whenever all it’s handles are closed.
I wonder why no installers use this?
That’s because it’s a lie. If you can’t delete a file in Windows, you can’t rename it either.
SYSTEM can do all of this. It works quite similarly to UNIX — ie. the whole unlinking while the file is opened.
Yes, but do you know what the behaviour of programs using said shared library will do when it’s changed out?
On a *nix system the old version is kept in memory until the last program using it closes, and new instances will load the new version.
Being able to do something, and it having defined behaviour are two different things.
And how many Windows users care about this kind of functionality? Not many.
Those that do are server admins, and they run Server 2003 w/ SP1, which provides hot-patching.
Actually, in some cases you can upgrade a kernel module without rebooting the system.
Not to mention the fact that almost all Linux software (including daemons) can be upgraded without a reboot.
The only things that require a reboot in Linux are updates to the core kernel and glibc. This is why users of highly stable distributions, such as Debian, can keep their system up to date for three years, with maybe three reboots during that time.
The only things that require a reboot in Linux are updates to the core kernel and glibc
No, actually, updating glibc does not require a reboot (I know because I do it). You can go back to single level if you don’t want to crash all your apps when removing the older version though.
Actually, in some cases you can upgrade a kernel module without rebooting the system.
In some cases? No. You can nearly always do that without rebooting the system. rmmod the module, compile the new version, and insmod it. Pretty straightforward…
glibc upgrade does not need a reboot, it needs some app restarts. I did it on my work box, and it worked fine. I only rebooted, because I had to restart X, because it crashed for some reason. And I worked on that box for ~160 days. So it was time for a kernel upgrade.
Have you heard of kexec? It’s in the mainstream Linux kernel already.
“Linux -> upgrade of kernel requires a reboot. Update of operating system (usually major revision jumpes i.e fc 3 to 4) requires reboot.”
Wow, man! An upgrade to kernel requires a reboot of the kernel?! Who would’ve thought! Next you’ll probably tell me that I’ll have to halt my computer before changing the PSU.
Pure BS. PSUs have been hot-swappable for decades.
Just not in most PCs.
😉
Applications demanding a kernel module do not require a reboot.
“but they all do it”
You’ve obviously never used Stratus VOS. I’ve never actually used it, but I first came across it in the mid 80s, and even then, the ONLY thing that forced a reboot was a UPS failure.
Even a kernel upgrade was NOT a “needs reboot” scenario.
Cheers,
Wol
Granted windows might need to do it a bit more than the others
Just a bit would make that one of the bigger understatements EVER.
“upgrade of kernel requires a reboot”
What OS doesn’t? And this only goes for the core kernel, virtually all (there are a few that don’t remove cleanly, but they’re pretty rare) kernel modules can be removed and replaced with a new version safely.
“Update of operating system (usually major revision jumpes i.e fc 3 to 4) requires reboot.”
Mostly because the kernel was upgraded. Also because there are probably new startup scripts that need to be run before some of the services can be restarted. And asking for a reboot when upgrading your OS and virtually every software package on the system is not unreasonable. The point is to avoid 10 reboots for a single installation (I’ve had to go through this installing or upgrading Windows before, it’s not fun).
most reboots in win 2k/xp are unnecessary. installer writers are lazy and request useless reboots.
but of course zealots have only very selective perseption of the world. btw, i would like to have something like an “reboot manager” on linux, since it is not always that obvious what to restart and how.
You already have one, it’s called /etc/init.d
Upgrade scripts already restart services that need to be restarted. You can restart them manually if you want. Even better, you can add your own services by writing a simple shell script. Windows probably requires a 100mb GUI program with 1,000 options that never works right. Exaggeration aside, there’s something to say for simplicity over dumbing-down for people who are too afriad to, or simply too stupid to, read and write shell scripts.
Actually, /etc/init.d is a directory that contains the startup and shutdown scripts. Your answer is akin to saying “/lib/kernel” as a response to a question of how to list kernel modules. As a long-time Unix user, I know of only one way to keep track of running services: “ps”.
If any Linux distribution includes a well-integrated daemon manager such as Windows does, I would like to know about it!
“ps” is a good command, however “ps -aux” is more useful
now a well integrated daemon manager that might work well for you is called GPS
http://www.tucows.com/preview/58425
it has processes set out by user / memory use /cpu use /host etc etc and gps gives more control over each process than Windows could ever wish for, kill, hangup, renice, POSIX signals, network watching
GPS is one of them apps that you wonder how you managed before you used it.
Gah!
For linux you have some tool (dont remember the name) that lets you change even modules of your kernel without rb.
http://www-128.ibm.com/developerworks/linux/library/l-kexec.html
04 May 2004
Even if your work doesn’t require you to reboot your Linux machine several times a day, waiting for a system to reboot can be a real drag. Enter kexec. Essentially, kexec is a fast reboot feature that lets you reboot to a new Linux kernel — without having to go through a bootloader. Faster reboot is a benefit even when uptime isn’t mission-critical — and a lifesaver for kernel and system software developers who need to reboot their machines several times a day. Kexec is currently available on the x86 32-bit platform only.
Wow.
wow32.
wow64.
Is there another OS that does not do this yet for years?
.V
They’re doing their best to catch up with Unix, perhaps they will succeed … one day
Well done guys !!
Finally they manage (it’s beta so maybe it’s fake
) to copy Unix !! Super 30 years delay !!
Maybe we need 10 GB of RAM to use it or 6 PentiumV 40Ghz ?! Start making a banking plan to buy a PC for Vista…
So incredible ! What is the worst is that 80% of people using windows will think they have invented this !!
A huge exaggeration.
Alot of the time application developers are at fault rather than windows. 9/10 when an app installer demands a reboot you don’t actually need to do it.
Like I said they all do it. Theres no arguing windows does it more than others and any reduction is good news but from the posts so far anyone would think unix systems (with the possible exception of ‘proper unix’ like solaris, aix etc) never ever require a reboot when being updated.
The day linux et al can have the kernel reloaded on the fly then you can climb on your high horse (and I’ll be right there with you, if that ever happens it will be fantastic) but until then recognise that this is a ‘flaw’ with all operating systems to some degree.
“A huge exaggeration.
Alot of the time application developers are at fault rather than windows. 9/10 when an app installer demands a reboot you don’t actually need to do it.”
You are of course 100% right. Many there are many times a installer of an application and even microsoft patches have demanded a reboot.
However if you ignore the reboot request you will find that the application is fully functional without a rebooting.
> A huge exaggeration.
Hello… It’s the internet.
> Alot of the time application developers are at fault rather than windows.
> 9/10 when an app installer demands a reboot you don’t actually need to do it.
Well, if Linux can be blamed for a rare PHP vulnerability (“Is This the Dawn of the Linux Worms?” — http://www.osnews.com/story.php?news_id=12601 ), then Windows can be blamed for every little hang-nail reboot.
I guess you’ve never heard of kexec mr. smarty pants. Linux can reboot “on the fly” now this will bring down all services, but the machine won’t go through any bios self test.
It is more of a “switch to single user mode, load new kernel, restart init,” kind of thing. That is still LIGHTYEARS ahead of anything Windows will be able to do for the forseeable future.
I change the name of my windows, i need to reboot.
I install active directory with dcpromo, i need to reboot.
So, Microsoft developers SUX!
Alot of the time application developers are at fault rather than windows. 9/10 when an app installer demands a reboot you don’t actually need to do it.
You do include Microsoft themselves in that group of 9, right?
from the posts so far anyone would think unix systems (with the possible exception of ‘proper unix’ like solaris, aix etc) never ever require a reboot when being updated.
Only a major windows fanboy could come to that conclusion. From your posts I would draw that UNIX is just as bad as windows because it can’t replace the very kernel on the fly without booting the new one and that’s just a wild delusion that could only come from somebody who never actually managed UNIX systems of any flavor… or a paid ms shill.
“The day linux et al can have the kernel reloaded on the fly then you can climb on your high horse (and I’ll be right there with you, if that ever happens it will be fantastic) but until then recognise that this is a ‘flaw’ with all operating systems to some degree.”
I partially agree with you on this. I’m not sure it is a flaw in the OS, but also linked to the entire architechture of computers (read: x86 arch) as we know it.
When we boot a machine the first thing we encounter is the BIOS, it then detects basic hardware and the the actual OS i loaded.
In order for a new OS (kernel upgrade or entire new OS) we need to revert to this state.
If we can get to this state without a “reboot”, i.e., shutting down all applications, we are a step closer. BIOS’less machines could be a step closer. I hope they are.
“The day linux et al can have the kernel reloaded on the fly then you can climb on your high horse”
Have you heard about kexec? It’s still experimental, but some kernels have it, and it allows to load a new kernel over the existing one, effetively replacing the running kernel with a new one without a reboot. Well I’m running a Debian server for a small company as Samba & mail server, it’s running for over a year now, the only reboot was because we had to install new power cables. I’ve made endless updates so far, from applications (samba, postfix etc.) to device drivers to libraries and I didn’t have a single reboot. Okay, I would have had to reboot if I had done a kernel update, but that was not necessary so far.
Tom
What is with the pessimistic comments here…would you prefer that microsoft leave things the way they were so you can continue to whinge about them? Yeah Yeah they were well behind on this front but then again they are ahead on quite a few as well.
Good on them for fixing up what would have to be one of my biggest gripes with windows…now about that registry.
But they’re fixing the effect, not the cause.
No, we don’t want Microsoft to leave things as they are. We want Microsoft to go to hell and take Windows and all the shit they produce with them too.
>=> on windows you reboot on :
>- driver upgrade
Have fun upgrading any driver on linux that isn’t a kernel module with rebooting. And they do exist.
>- software upgrade
See my note above, more the app developers fault. THe only time you usually *have* to reboot in reality when installing an app is when it installs a device driver. See my first point above.
>- freeze of system
Hate to break it to you, linux freezes, OS X freezes hell there is a bug somewhere in the freebsd 5 geom.bde and / or file loopback drivers that means you can lock up a freebsd system at will.
WHats your point? again all operating systems do this. Windows maybe more or less…but once again the incoherent rantings already happening in this thread make it sound like unix systems never fail, never require a reboot and are generally without fault.
“Have fun upgrading any driver on linux that isn’t a kernel module with rebooting. And they do exist.”
Luckily the only drivers i regularly upgrade (regularly = 1 every 6 months approx.) are kernel modules =)
“See my note above, more the app developers fault. THe only time you usually *have* to reboot in reality when installing an app is when it installs a device driver. See my first point above. ”
Would those app developers include microsoft themselves. Windows Updates need restarts even though a lot of their apps dont install device drivers (and they bug me about it to). Restarting may not be required, but it sure as hell is recommended to maintain system stability of not only the new application, but also the OS itself. Dont try to downplay what MS is doing with Vista. It is a big deal and congrats to MS for finally moving foward. I think OSX should try to reduce recommended system rebooting as well after their updates
Using various *nix and windows and MACOSX i know that *nix users can confidently tout how they have to reboot their machine much less.
“Luckily the only drivers i regularly upgrade (regularly = 1 every 6 months approx.) are kernel modules =)”
If the kernel had some more closely resembling a stable set of KPIs, you wouldn’t even need to do that anywhere near as often as you do now. While its true that kernel interfaces are going to change over time, Linus and gang are hell bent on making a point of it by changing interfaces on purpose (Google the LKML archives for proof of that statement).
“Would those app developers include microsoft themselves.”
Yes actually, and in my experience, they are in fact the worst offenders as most Microsoft applications come with updated versions of core DLLs, and it would be a mess to install their software and not reboot. A similar situation exists on Mac OS X, but to a lesser extent.
“Dont try to downplay what MS is doing with Vista. It is a big deal and congrats to MS for finally moving foward. I think OSX should try to reduce recommended system rebooting as well after their updates”
I whole-heartedly agree on these points.
“Using various *nix and windows and MACOSX i know that *nix users can confidently tout how they have to reboot their machine much less.”
That’s my experience as well.
Yes actually, and in my experience, they are in fact the worst offenders as most Microsoft applications come with updated versions of core DLLs, and it would be a mess to install their software and not reboot. A similar situation exists on Mac OS X, but to a lesser extent.
Well, there definately isn’t the problem with MacOS X as I can see; when was the last time you saw an application unilaterally over right system libraries with their own ‘custom’ version?
If Microsoft wish to solve all the problems associated with Windows; make it impossible to access the Windows directory, unless updating it with a service pack or updaters from Windows Update – there should be absolutely NO reason for for ANY application vendor to write ANY dlls to the Windows directory.
If these companies find that they MUST include their customised/updated version of a Windows dll; then why don’t they simply include it in their own application directory? atleast everything would be located in one directory, like how Apple’s .app system works, with everything shoved under one fat directory.
If Microsoft wish to solve all the problems associated with Windows; make it impossible to access the Windows directory, unless updating it with a service pack or updaters from Windows Update – there should be absolutely NO reason for for ANY application vendor to write ANY dlls to the Windows directory.
They call that Windows XP.
Incorrect; you can still write DLLs to the Windows directory – cutting short the idea of executing people who write DLLs to Windows directory, the next best option is to make it completely unaccessible.
You can write them there but the system won’t let you overwrite the essential ones. The only argument in this case is other DLL’s such as those used by the non-core windows system. I’m not arguing that, just that not everything is a DLL overwrite free for all. I hate XP.
It most certainly does let you overwrite the critical Windows DLLs. When you do it, you get a dialog box that pops up and asks “do you want to keep the unrecognized versions?” and you click OK and it’s done. MS put a dialog box with an OK button right before the potentially devestaing action. The same excellent security model they’ve been pursuing for the last 10 years, and has worked so well to secure IE. I can definitely understand why all you kids are so crazy about Windows.
But why even PLACE them in the Windows directory in the first place? just install the bloody DLLs in the applications directory, along with everything else! Maybe execution is the only way to motivate programmers to keep all their crap in one directory rather than spralling it from one end of the hard disk to another.
that has already been done with windows 2000.
If you overwrite a system file it gets replaced by the original file via system file protection.
If the kernel had some more closely resembling a stable set of KPIs, you wouldn’t even need to do that anywhere near as often as you do now. While its true that kernel interfaces are going to change over time, Linus and gang are hell bent on making a point of it by changing interfaces on purpose (Google the LKML archives for proof of that statement).
What does that have to do with what he said;
“Luckily the only drivers i regularly upgrade (regularly = 1 every 6 months approx.) are kernel modules =)”
…as in, if he updates a kernel module, it is loaded on a running system. No restarts.
Umm when i said i regularly upgrade kernel modules, that means i dont restart. I just reload the modules
The only thing that requires a restart is when i do a kernel compile……..which i do like once every 5 years (im being generous here) or so. Im still on kernel 2.4 =). In fact since im using slackware all i did was take out the kernel packages, compile my own kernel with my specifications, and installed that kernel. Thats about it. I wont upgrade to 2.6 until linux moves on to 2.8.
Honestly the only time I reboot a linux box is when my paranoia tells me too because it might potentially explode after running it for 18 months or so =) =). I love my computer and I dont like to punish it =).
Mr Windows fanboy, I suggest you stick to commenting on something you might know something about.
Have fun upgrading any driver on linux that isn’t a kernel module with rebooting. And they do exist.
This doesn’t even make sense, you maybe want to rethink your sentence. If you mean without rebooting, well, upgrading a driver that isn’t available as a module means you’ve upgraded or patched your kernel. Of course you have to reboot. Duh.
See my note above, more the app developers fault. THe only time you usually *have* to reboot in reality when installing an app is when it installs a device driver.
What? Heard of insmod or modprobe?
Hate to break it to you, linux freezes, OS X freezes hell there is a bug somewhere in the freebsd 5 geom.bde and / or file loopback drivers that means you can lock up a freebsd system at will.
Now, I can’t speak for OSX or FreeBSD, but I don’t think I’ve ever seen a linux system “freeze”. I’ve seen X freeze for various reasons, but if you use ssh or something similar the system is still alive. You kill X and go on with you business. Now, for a windows user the difference might be moot, but the fact is that this is a huge deal, especially if you run some sort of server daemon on the system.
WHats your point? again all operating systems do this. Windows maybe more or less…but once again the incoherent rantings already happening in this thread make it sound like unix systems never fail, never require a reboot and are generally without fault.
Windows DOES tend to fail a lot more often, in my experience. Part of this is probably due to running graphics drivers in ring 0. Almost all crashes/freezes I’ve seen with windows, except for 3.x and 9x, has been related to bad graphics drivers. The rest has been related to a confused or damaged registry. Considering that graphics-drivers are usually not hysterically optimized for performance or running in ring 0, *and* the absence of a braindead registery, it’s hardly surprising people have a better experience with unix-based systems.
Again, checkout kexec in the mainstream Linux kernel. Also when Linux does freeze, something is seriously wrong. My Gentoo system has hardlocked only three times on one single week in it’s entire three year existance. I tracked the bug down that week to a change in the crypto-loop driver, been fine since.
And how is that different from Windows? When Windows (9x series excluded) freezes, it is 98% of the time a bad driver.
“Have fun upgrading any driver on linux that isn’t a kernel module with rebooting. And they do exist.”
The only propietary driver I use is for my NVIDIA card, and if I update it, I do a “modprobe -r nvidia; modprobe nvidia”, hit Ctrl+Alt+Backspace which quickly restarts X, which doesn’t even take a few seconds, and I’m back where I’ve been without rebooting the system.
Tom
Is it just me or is every comment posted instantly given a +1 score?
They’ve implemented their latest version of OSNews. Registered users start at +1, among other minor changes.
‘Which part is the exaggeration?’
{“Upgrade of Kernel requires reboot”, while Windows needs it for, err, almost everytime something is installed or settings changed!
(note, slight exaggeration)}
All of that? Changing my desktop settings doesn’t require a reboot. The overwhelming majority of apps don’t require a reboot (even if they ask for it).
“Windows Updates need restarts even though a lot of their apps dont install device drivers (and they bug me about it to)”
I’d equate windows updates to, roughly, an equivalent of a linux kernel recompile or at least a freebsd makeworld. They (usually) affect the significant core areas of the operating system.
Again…any OS with operating system patches usually requires a reboot (again exluding the big vendors like solaris etc)
“Restarting may not be required, but it sure as hell is recommended to maintain system stability of not only the new application, but also the OS itself.”
Not necessarily. I’ve had apps which I *know* have only dumped files into a folder, created a desktop shortcut and kept all their settings in a .ini not the registry..ask me for a restart. Why?
Also games are notorious. ‘Would you like to install directx?’ no..I have the latest version. ‘Install canno t proceed until you install directx’. ok..install it then. ‘we have checked your version of directx and you do not require an update..but I’m going to get you to reboot at the end of the installer anyway just in case’. doh.
“Dont try to downplay what MS is doing with Vista. It is a big deal and congrats to MS for finally moving foward”
woah there. I’m not, I think its great. Anything on any operating system that mitigates the need to restart is fantastic, and I’m excited about this particular feature. I hope it gets included and works as well as or better than advertised!
What I’m taking offense to is flippant comments like ‘welcome to the 20th century’ and other such diatribe when, in fact, all operating systems have some kind of reboot after installing OS updated / applications problem. I look forward to the day when the linux kernel can be recompiled and reloaded on the fly, ditto for bsd’s.
“Using various *nix and windows and MACOSX i know that *nix users can confidently tout how they have to reboot their machine much less.”
Again, comes back to if its actually required and general usage patterns. You’ve already said you upgrade a driver that requires a reboot every 1 to 6 months. Me too under windows..the nvidia drivers. Other than that I rarely reboot. I lost a my main desktops windows uptime of around 3 months (and that began because it was a new machine) last week because a dvd write failed and the disk was soft locked in the drive. Had to reboot to eject it. This is a machine I use for a bit of everything, web browsing, email, work, games, films, audio etc
Windows’ fault or the applications fault?
Most apps function correctly if they havn’t installed a device driver, or at least well enough that I don’t notice until I *do* have to reboot.
Also theres the general server vs desktop argument, but I’m going in to unknown terriotry here. I’d be genuiinely interest in some kindof comparison between reboot rates between similar desktops under linux and windows. Say two stable machines running in a corporate environment. How the reboot rate due to updates / upgrades is affected.
So yeah, great feature but it’s not ‘catching up’ any to linux et al in my view. They’re all on the same level.
I’d equate windows updates to, roughly, an equivalent of a linux kernel recompile
Give me a break. A patch for IE or for Windows Media is no way near a kernel recompile.
You try to equate everything with a kernel recompile on Linux because that’s the only thing you think require a reboot on Linux. Sorry to tell you that even a kernel recompile on Linux does not require a reboot. It does require one to take effect, but you’re not required to reboot.
They (usually) affect the significant core areas of the operating system
Are you saying that Windows kernel is riddled with bugs ?
necessarily. I’ve had apps which I *know* have only dumped files into a folder, created a desktop shortcut and kept all their settings in a .ini not the registry..ask me for a restart. Why?
Don’t know, but the fact is they require a reboot in Windows.
thing on any operating system that mitigates the need to restart is fantastic, and I’m excited about this particular feature
Especially since your “any operating system” actually refers to Windows. The fact that you are excited by this is proof. This is several decades late it’s not funny.
What I’m taking offense to is flippant comments like ‘welcome to the 20th century’ and other such diatribe
You mostly take offense because it’s true.
all operating systems have some kind of reboot after installing OS updated / applications problem
No, they don’t and you’re wrong. Only Windows has this problem.
I look forward to the day when the linux kernel can be recompiled and reloaded on the fly
Stop your BS, what you say here is not in the same league as what MS is trying to do.
Actually, latest Linux kernels can already do that, to load a minimum kernel and dump memory to disk, so you’re really very late.
You’ve already said you upgrade a driver that requires a reboot every 1 to 6 months
As a matter of fact, if you loaded your drivers as module, you can reload them on the fly already.
You can’t for some that can’t be unloaded because the OS rely on them (like disk drivers).
Changing my NVidia binary driver do NOT require me to reboot for example.
Windows’ fault or the applications fault?
Windows’ fault. If my CD is locked in my drive, sure enough, I will accuse Linux, not K3B.
But now again, I can always eject discs on my Linux box …
Most apps function correctly if they havn’t installed a device driver, or at least well enough that I don’t notice until I *do* have to reboot
Given that you don’t know which ones work well in this condition, the default behaviour is to reboot in Windows to be sure (Windows puker at work). In Linux, the default behaviour is not to reboot anything. Do you understand the difference ?
Say two stable machines running in a corporate environment. How the reboot rate due to updates / upgrades is affected.
I already told you. Every Linux desktop distro provides auto updates, and they never require a reboot.
So yeah, great feature but it’s not ‘catching up’ any to linux et al in my view. They’re all on the same level.
Of course MS is catching up to Linux, but you’re so blinded by zealotry that you can’t believe the truth when it is right before your eyes, hence your view.
Sure enough Linux and Windows are not in the same level.
‘Mr Windows fanboy, I suggest you stick to commenting on something you might know something about. ‘
I’m not a fanboy. I run windows on my desktop and linux, solaris, freebsd and OS X in mixed server and desktop roles. Best tool for the job.
‘This doesn’t even make sense, you maybe want to rethink your sentence. If you mean without rebooting, well, upgrading a driver that isn’t available as a module means you’ve upgraded or patched your kernel. Of course you have to reboot. Duh.’
ANd what difference is there to installing a driver under windows which requires a reboot?
Bottom line : under windows you have to reboot for drivers
under linux this is still an occurance. It may be slightly rarer but it does happen.
‘What? Heard of insmod or modprobe? ‘
See above? This isn’t always possible.
‘Now, I can’t speak for OSX or FreeBSD, but I don’t think I’ve ever seen a linux system “freeze”‘
Good for you. I have, many times.
‘Windows DOES tend to fail a lot more often, in my experience. ‘
I’m not arguing particularly against that point. All I’m saying is, in my view its not significantly worse a problem than with any other OS (with the caveat that you have to know what you’re doing which I don’t think many people do no matter how computer literate they think they are)
My real beef, like I’ve said, is the taunts about windows catching up to linux etc on this front. When in fact its a problem that plagues them all to some extent.
Ok, first of all, would you please consider using “reply”? Thank you.
ANd what difference is there to installing a driver under windows which requires a reboot?
The difference is that drivers in windows are *modules*, ie the variant in linux that *doesn’t* require a reboot. You remember the Windows uses a microkernel propaganda? One of the supposed benefits of that is that changing a driver would *not* require a reboot, as opposed to linux which is a monolithic kernel that has to have the most basic drivers compiled in.
Bottom line : under windows you have to reboot for drivers, under linux this is still an occurance. It may be slightly rarer but it does happen.
This isn’t really the point, the point is that windows or applications wants to reboot all the time for trivial reasons, like for instance installing some sort of cdrecording software. This sw probably adds some driver file somewhere that doesn’t get used/found until windows is rebooted. As a linux user you just scratch your head at this one.
So the bottom line is actually, upgrade or install ANY driver in windows and you’ll have to reboot, as oposed to linux where you only have to reboot if you upgrade or change some really fundamental core part of the kernel. Adding some software that adds some driver does not require a reboot, unless the application requires you to patch your kernel sources, compile and update your kernel.. In windows you usually call that “upgrading to a new version of the OS”.
Drivers that isn’t a vital part of system doesn’t require a reboot in linux. You are equating changing drivers for the sound system, or your nic, with switching drivers for your disc controller or boot/root file-system driver or something else truely fundamental.
See above? This isn’t always possible.
It would be very helpful if you could specify when it’s not. Perhaps you are afraid to get refuted? Basically the only drivers you can’t upgrade without a complete reboot are those that controls *vital* parts of the system, like filesystem drivers disc controllers and such. However, this would rarely occur since you are much more likely to upgrade the entire kernel than upgrading specific drivers.
Good for you. I have, many times.
Nobody here knows what you consider a “frozen” system. For all we know you could mean a system that no longer heads input from keyboard/mouse.
I’m not arguing particularly against that point. All I’m saying is, in my view its not significantly worse a problem than with any other OS (with the caveat that you have to know what you’re doing which I don’t think many people do no matter how computer literate they think they are)
Well, you actually are, since you blow off *the* two most frequent causes of problems with windows, of which
neither are present in the other systems by saying there is no significant differnce.
My real beef, like I’ve said, is the taunts about windows catching up to linux etc on this front. When in fact its a problem that plagues them all to some extent.
See what I and others have written above about what updates forces you to reboot.
My real beef, like I’ve said, is the taunts about windows catching up to linux etc on this front. When in fact its a problem that plagues them all to some extent
You are some powerful zealot.
With the same extent as you, I could say that “every hardware works with Linux to some extent”.
I would not be more wrong than you.
Windows does load drivers dynamically. I’m a storage guy and we always see Server 2003 boxes not ask for reboots when adding and removing software devices dynamically. Actually if you use HP hardware and install the Proliant Support Pack it will upgrade almost all hardware drivers without a reboot, HBA’s NIC’s RAID and Video drivers. Most of the time when you need to to reboot Windows Server it is because you loaded an app with a File System filter driver like a Virus Scanner software package. To get many of theese drivers to work properly you have have the driver loaded before mounting the volume the problem is nine times out of ten you can’t unmount and remount a volume on an active box so most manufactures just say reboot to ensure you remove and exclusive locks on the box and the file system driver then loads before the volume mounts on the reboot.
And yes most applications are cross windows version so the installer just assumes it needs to reboot the box because Windows 98 may have needed but XP doesn’t.
‘Hello… It’s the internet. ‘
hah true…although this place is even more rabid than slashdot. Something I never thought I’d say!
‘And yes most applications are cross windows version so the installer just assumes it needs to reboot the box because Windows 98 may have needed but XP doesn’t.’
True but this doesn’t make it windows’ fault. Rather the installer should enumerate the version of windows running and adapt accordingly.
Your storage comments were very interesting. I only maintain one 2k3 server with fairly static hardware so I’m not familiar with how it handles things.
We consantly add new luns to servers while online but the real kicker is having tape devices and changers come online to dozens of hosts at the same time because we had to reboot a bridge or added a new library. Drivers add and upgrade with no intervention most of the time. Truth of the matter Windows is far easier than most *nix in a SAN envronment because it’s not dependant on special files and hardware paths.
The day linux et al can have the kernel reloaded on the fly then you can climb on your high horse (and I’ll be right there with you, if that ever happens it will be fantastic) but until then recognise that this is a ‘flaw’ with all operating systems to some degree.
If I’m not mistaken, some IBM mainframes running AIX can do that…
Linux has taken a small step in that direction with kexec. kexec allows you to boot a new kernel without rebooting or using a bootloader. It still requires all processes to be stopped, so it’s not quite “on-the-fly” replacing of the kernel.
See http://www-128.ibm.com/developerworks/linux/library/l-kexec.html
Way to go Microsoft! Lead the pack for innovation!
Linux has only had this feature since… well… Linux.
I only reboot my servers when I install a new kernel and
even then I wait until it is convenient most of the time.
‘If I’m not mistaken, some IBM mainframes running AIX can do that… ‘
I’m not sure on specifics but yes, its probably likely.
Sorry all through my posts I’ve been adding the caveat that the ‘big iron’ unix systems like Solaris, aix, HPUX etc probably differ in behaviour in this regard…but they aren’t really the focus of this web site. I should have also added that disclaimer to that post.
‘See http://www-128.ibm.com/developerworks/linux/library/l-kexec.html‘
I hadn’t read about that before. Thats pretty interesting. I can’t see the new vista features coming up to this level but I’ll happily be surprised on that front
‘Again, checkout kexec in the mainstream Linux kernel.’
I’ve recently been enlightened
However, as the initial poster stated this is pretty much still a reboot in all but name. I will assume that it happens far more quickly than you general boot procedure though (no bios check, bootloader and likely less disk access as most process could still be in memory cache).
‘Also when Linux does freeze, something is seriously wrong. ‘
Straying away from the main argument here…but the same can be said for windows. The only person attempting so far in the thread to explain why windows falls over more often has used graphics drivers as an example. Is this the fault of windows (ring 0 as mentioned although linux drivers are also in the kernel space which makes them almost as bad) or the driver?
‘ I tracked the bug down that week to a change in the crypto-loop driver, been fine since.’
Until the next bug anyway….
Not ragging on linux, these things are inevitable. Like I said I had a freebsd 5.4 machine with an encrypted loop back device that I could crash at will. There have been changes in the geom code which may alleviate it but I’ve yet to test.
‘Wow, man! An upgrade to kernel requires a reboot of the kernel?! Who would’ve thought! Next you’ll probably tell me that I’ll have to halt my computer before changing the PSU.’
Hi, welcome to the thread. Thanks for proving my point.
As an aside…all those touting kernel modules as the solution to the argument many machines I’ve admin’d and for sure the ones I set up myself have kernel modules disabled in the kernel for security reasons. Particulary for machines in ‘high risk’ situations (unfriendly network, handling financial or medical records).
Linux graphics drivers are decidedly NOT in kernelspace, unless you are talking about either propreitary binary drivers for nvidia/ati, or the framebuffer drivers used to get graphics for the console. You seem to be seriously misguided.
‘I guess you’ve never heard of kexec mr. smarty pants’
Actually at the time I read your post I had
‘It is more of a “switch to single user mode, load new kernel, restart init,” kind of thing. That is still LIGHTYEARS ahead of anything Windows will be able to do for the forseeable future.’
As has been pointed out its still pretty much a restart in all but name, but hopefully a speedy one.
As for being lightyears ahead of windows did you read the summary above mr smarty pants? Theres no guarantee this *won’t* occur in vista.
Unlikely I’d say but you never know.
Also I doubt many people use kexec day to day. It certainly took ~30 comments here before someone brought it up.
Is kexec stable? Id be surprised if it is anywhere near stable.
‘Applications demanding a kernel module do not require a reboot.’
Where did I say they did?
Do people here actually read these comments?
However, in the (not as rare as you might think) instances where :
a) kernel modules are not available
b) the kernel does not allow modules to be loaded
You would have to reboot.
All this is still getting off the original point I was rebutting.
Stop acting like linux is some holy ‘doesn’t need rebooting ever, has no problems’ operating system.
I believe I initially piped up to respond to the ‘welcome to the 20th century’ comment.
Which implies that linux is sooo far ahead of windows in this area.
When, in fact :
*both* require rebooting in certain situations when installing a device driver
*both* require rebooting for upgrades to the core OS
*both* require rebooting for patches to the OS
*both* require rebooting during certain app installs
I’m not arguing whether one is rebooted 10 times in a year whereas the other is only rebooted 6. The numbers are moot.
The fact is most operating systems are afflicted with this and it tires me to hear the linux evangelists get on their high horse regarding this area.
Do people here actually read these comments?
Do you read them yourself ? You still dod not understand what reality is.
I’ll help you there.
Stop acting like linux is some holy ‘doesn’t need rebooting ever, has no problems’ operating system
It’s not, but that’s because you just made a straw man. Nobody said that, only you.
I believe I initially piped up to respond to the ‘welcome to the 20th century’ comment
Which may offend you, but is still the truth. Actually, MS is still not in the 20th century with its FS that still heavily fragment.
Which implies that linux is sooo far ahead of windows in this area
Of course it is, on ly a zealot can not see that.
*both* require rebooting in certain situations when installing a device driver
Wrong. Not in Linux.
*both* require rebooting for upgrades to the core OS
Wrong. Not in Linux.
*both* require rebooting for patches to the OS
Wrong. Not in Linux.
*both* require rebooting during certain app installs
Wrong. Not in Linux.
You are wrong on all counts. The only thing that require a reboot is change of kernel on Linux.
I’m not arguing whether one is rebooted 10 times in a year whereas the other is only rebooted 6. The numbers are moot.
Your numbers are moot. In most cases, the numbers would rather be 50 times for Windows against 2 times for Linux, which are not so moot numbers.
The fact is most operating systems are afflicted with this and it tires me to hear the linux evangelists get on their high horse regarding this area
Not they are not but you are such a zealot you can’t accept the truth.
I recently did a group test of a variety of media player apps on a secondary PC I use for testing software. About half of them required a reboot when they were installed and a couple required another reboot when they were uninstalled. In the past I’ve had simple image/text editors, encyclopedias, utilities, games and loads of other apps require a reboot. It’s a real pain when the computer is being used for other tasks.
It’s all very well saying that some of them don’t really need the system to be rebooted, but how am I meant to know which they are? One time I tried to use an encyclopedia after installing it rather than rebooting first. The app crashed with a load of error messages and after that it wouldn’t uninstall cleanly and caused registry problems.
This on it’s own isn’t such a big deal that I’d switch to Linux because of it, that OS has plenty of annoying problems of it’s own, but it’s very nice to see this flaw finally get fixed.
‘You do include Microsoft themselves in that group of 9, right?’
Yes..but I’m not sure what that has to do with anything?
Specifically as to why it would be windows’ fault.
Microsoft has different teams for different projects. I’m sure from the outside it would make sense to say they should all be intricately aware of each others work / collaborate tightly but I’m pretty sure…as is the case with any large company…this doesn’t happen as frequently as it should.
BeOS had this in 1996 or 10 years ago.
No need to reboot at all only restart the your different daemons and you’re done.
I was thinking the same think !!!
I’m speechless.
‘You’ve obviously never used Stratus VOS. I’ve never actually used it, but I first came across it in the mid 80s, and even then, the ONLY thing that forced a reboot was a UPS failure.
Even a kernel upgrade was NOT a “needs reboot” scenario. ‘
Again all through my comments I’ve liberally sprinkled the caveat that bigger unix systems like solaris, aix aren’t nececessarily afflicted by this.
However, osnews seems primarily to deal with the more consumer grade unix variants (specifically linux, *bsd, OS X and more and more on solaris now it is ‘open’).
I’ve never heard of Stratus VOS and, as you pointed out, certainly never used it.
And certainly most of the arguments have been in relation to linux vs windows.
But I believe I covered my back regarding this with the comments I’ve previously left. I couldn’t find a stratus VOS topic section in the OS News directory.
I have just read top ten comments and I have enough…
I’m a long-time OSNews reader.
It is a pity to see how this site is “slashdotting” itself day by day..
David
‘Give me a break. A patch for IE or for Windows Media is no way near a kernel recompile. ‘
Fair point, but then again not all windows updates are for these things.
‘You try to equate everything with a kernel recompile on Linux because that’s the only thing you think require a reboot on Linux.’
I’;ve also pointed out you’d have to reboot linux for large operating system upgrades (ala fc 3 -> 4). And in the case of freezes and sometimes in the case of software installs (although this is usually linked to kernel manipulation but could just as easily require a core library update etc)
‘Sorry to tell you that even a kernel recompile on Linux does not require a reboot. It does require one to take effect, but you’re not required to reboot. ‘
So..err…why bother compiling it in the first place?
I counter your point with if I run a windows update I don’t have to reboot and if I install an app even if it needs to reboot to function I don’t have to reboot.
Bit of a silly thing to say.
‘Are you saying that Windows kernel is riddled with bugs ? ‘
I’m not sure riddled would be an apt description. But it has bugs yes. All operating system do. Kernel release contain bug fixes as well as new features you know.
‘No, they don’t and you’re wrong. Only Windows has this problem. ‘
So you’ve never rebooted your linux machine due to an update (kernel, general operating system / userland or app) ever?
‘Don’t know, but the fact is they require a reboot in Windows. ‘
Read what I said. They do not *require* one. They ask you to reboot. If you decline it does not affect their runtime behaviour. Ergo is that windows’ faiult or the app developers just covering their backs?
‘Especially since your “any operating system” actually refers to Windows. The fact that you are excited by this is proof. This is several decades late it’s not funny.’
Sorry? I also said I will be ecstatic the day linux can dynamically reload its kernel without a reboot (this is subtly different to what kexec seems to do). So no, my ‘any operating system’ comment refers to any operating system that does not require downtime for maintenance.
‘You mostly take offense because it’s true. ‘
Clearly a matter of opinion. I believe it’s not true, which I’ve attempted to quantify.
‘Stop your BS, what you say here is not in the same league as what MS is trying to do. ‘
I didn’t say it was, I was merely pointing out that any improvement that removes downtime is good.
I don’t think vista will do the above, but I hope it will in the back of my head. I also hope linux gets this. And bsd.
‘Actually, latest Linux kernels can already do that, to load a minimum kernel and dump memory to disk, so you’re really very late. ‘
Do you have a url? I’d very interested in this for the servers I run linux on.
‘Changing my NVidia binary driver do NOT require me to reboot for example. ‘
I know.
‘Windows’ fault. If my CD is locked in my drive, sure enough, I will accuse Linux, not K3B. ‘
Regardless of what you would or would not do…is that an accurate statement? What locked the drive the linux device driver or the burning application?
‘Given that you don’t know which ones work well in this condition, the default behaviour is to reboot in Windows to be sure (Windows puker at work).’
Here I can reference you to my post stating that somepeople don’t understand what they’re doing even when they think they do. Its possible to trace exactly what a program does at install time (and is generally good practice) and so you should be aware if something has altered windows in a way that you deem necessary for a reboot. So your ‘windows puker’ is actually someone who doesn’t know what they’re doing.
‘Do you understand the difference ?’
Yes, do you?
‘I already told you. Every Linux desktop distro provides auto updates, and they never require a reboot. ‘
You did not tell me. WHere are your url’s to studies in the matter?
They never require a reboot? Really?
My ubuntu ‘desktop linux distro’ wants me to upgrade the kernel image it’s using. You’re telling me that wouldn’t require a reboot? (in addition other updates my for some reason rely on that update).
Of course this goes back to your crazy run the update but never actually use it argument.
‘Of course MS is catching up to Linux, but you’re so blinded by zealotry that you can’t believe the truth when it is right before your eyes, hence your view. ‘
Why am I a zealout?
zeal·ot – A fanatically committed person.
I run linux every day as well as windows, which I’ve already mentioned. I’m committed to whatever does the job best for the task that needs done.
Shrug you keep telling me you never have to reboot your linux box even when running updates. I don’t see how this is possible.
‘Sure enough Linux and Windows are not in the same level.’
I think they are (and believe me, I would think that would be a far bigger compliment to linux than the other way round given the clamouring for desktop respect in the linux community). Indeed we’ve had people with storage experience chipping in saying that they find windows easier to manage in that environment.
Of cours eyou can argue with them too..you know best.
Fair point, but then again not all windows updates are for these things
But they all bug you for a reboot every time, with a dialog that stay on front of every app.
It’s unbearable. So they all require a reboot. I could not use a system with an annoying popup every 10 minutes for even one hour.
I’;ve also pointed out you’d have to reboot linux for large operating system upgrades (ala fc 3 -> 4)
That’s just not true. The reboot required is because they usually update the kernel too.
sometimes in the case of software installs
No this is just not true.
So..err…why bother compiling it in the first place?
The same reason for which you can have several kernel compiled to choose from at boot.
I counter your point with if I run a windows update I don’t have to reboot and if I install an app even if it needs to reboot to function I don’t have to reboot
Sometimes true for the app, definitely wrong for Windows update, as I don’t know a lot of people that can stand an annoying popup for hours. I know I can’t.
So you’ve never rebooted your linux machine due to an update (kernel, general operating system / userland or app) ever?
Of course not. You know what, I’m mostly annoyed at my laptop server because when a new kernel comes with a fix for a remote vulnerability problem, I can’t find a way to make it crash or stop, after having installed the new kernel (this happened once at start of this year though). The thing will go on running no matter what. I always have to reboot it manually. Same for glibc update. I recently updated the very same machine to the latest 2.3.6, without even going single level, and the thing is still tagging along (of course it is, as long as I don’t remove the old glibc).
Read what I said. They do not *require* one. They ask you to reboot. If you decline it does not affect their runtime behaviour. Ergo is that windows’ faiult or the app developers just covering their backs?
Given that Windows update do it, which are done by MS which develop the OS, it’s definitely Windows fault.
Given that nobody has to do that on Linux, yes, it is definitely Windows fault.
I agree that having to click on an annoying popup that comes above all windows and interrupts you every 10 minutes is not “requiring”, but I dare anyone to support that for a long time and then say to me they were not required to reboot.
I didn’t say it was, I was merely pointing out that any improvement that removes downtime is good
No it’s not. News to you : when the Linux kernel crash, sth is seriously wrong, most of the time a hardware problem that requires repair or change, that the kernel can’t handle. In the same league, if you HAVE to update your kernel, sth is seriously wrong too, so you can’t trust the current kernel to do a good job at reloading the new one.
I know it’s hard to believe for you.
What you propose will NOT fix anything, as the problem lies in the fact that a required reboot on Linux means your current kernel is not trustworthy enough, and that’s him that has to do the reboot.
Do you have a url? I’d very interested in this for the servers I run linux on
Search for kdump in the lkml. IIRC it is integrated in the kernel 2.6.14 now, following the summit on crash dumps for Linux. It reloads a new kernel to record the old one.
Regardless of what you would or would not do…is that an accurate statement? What locked the drive the linux device driver or the burning application?
Of course it is. But I know what you mean : during burning. When things like that happened, that was caused by buggy hardware that, well, could crash the kernel. So if it’s not the kernel, it’s the hardware.
It can’t be the app fault. If the app can lock the drive, it should be able to unlock it too.
Again, I never had these problems, but burning in Linux is actually better than in Windows since at least 1998, for which they went till inventing Burnproof/free, because so many people failed burning in Windows.
Its possible to trace exactly what a program does at install time (and is generally good practice) and so you should be aware if something has altered windows in a way that you deem necessary for a reboot. So your ‘windows puker’ is actually someone who doesn’t know what they’re doing.
Which is everyone. When I install an app, I don’t want to trace everything it does.
It needs reboot or it doesn’t. I never asks myself if I need to reboot in Linux : I don’t, without the need to trace anything.
You did not tell me. WHere are your url’s to studies in the matter?
You need studies to prove that distro upgrades do not requires reboot ????!!!!
Try to learn what you are talking about then.
My ubuntu ‘desktop linux distro’ wants me to upgrade the kernel image it’s using. You’re telling me that wouldn’t require a reboot?
No it doesn’t, you should know if you really tested.
Shrug you keep telling me you never have to reboot your linux box even when running updates. I don’t see how this is possible.
That’s because you don’t actually run a Linux box then. I recently updated my glibc, all of KDE to 3.5, all of Gnome to 2.12, all of XOrg to 7-RC2 and never have to reboot, I was not even required to quit the simultaneous KDE, Gnome and XFCE sessions.
Indeed we’ve had people with storage experience chipping in saying that they find windows easier to manage in that environment
So they’re very bad then. Especially talking of storage.
Excuse me I have installed, manged and configured hundreds of terabytes of storage (disk and tape) in Linux, HP-UX, Windows, Tru-64, Solaris even a little VMS in many different customer environments with many different storage platforms. So far the two best at handling device reconfiguration in the SAN have been Windows and Tru-64. If you have ever worked on an enterprise array and dealt with hundreds of LUN’s in a LVM config and have device special files change on you knows UNIX storage can be a bear (especially HP-UX). For the most part windows just works. Now with storport and MPIO Windows is starting to feel much more solid and depndable and seeing much better performance. Is Windows perfect no, is it getting better yes. Do some of the *NIX boxes work better in some cases? Yes they do. Do I care which one I work on, not really unless it’s HP-UX then I cringe but thats really because SAN’s on HP-UX are quirky.
At least I can use almost all my DOS, Win95, Win98,WinMe, Win2000 on my windows XP without re-compilation.
How much time will it take to recompile all the thing? Probably alot more that a reboot.
And btw, my windows do not crash. Maybe I’m god or something.
Oh, and my windows all have HW acceleration! Woo-hoo. Linux, will have in the future something like this.
And all my hardware works…
Well, I guess I will need to stay with my primitive Operation System that does requirea reboot from time to time.
If windows is catching up linux, well… you should say “welcome to the primitive age”, and not that “welcome to 21th century”!
“I believe I initially piped up to respond to the ‘welcome to the 20th century’ comment.
Which implies that linux is sooo far ahead of windows in this area. ”
I said that, Never mentioned Linux, it just every other OS I’ve ever used, MacOS, Amiga OS, QNX has never been so dependant on rebooting.
Remember the good old Windows 98 days, what’s the first thing you would do if something was working right? log off, and if that didn’t work, Reboot the PC.
Windows *has* got a lot better since then, and this seems to a step in the right direction, but still no attempt to fix the causes, just patch the problem… Which given how LONG it’s taken for Vista to come out, is rather disapointing..
Well, congrats to Microsoft for catching up with the rest of the computer world.
The concept of reboot part of a system has been around for quite awhile, I mean, the concpet is easy. Just look at any os the uses a *TURE* micro kernel, where the file system, drives, etc are all in userland. If Vista was based on a micro kernel, it would be easy to achive, but I don’t think Microsoft can really pull it off. With the hardware requirements, useing a micro kernel would requre even higher hardware standers. It is great for MS to dream big, but in reallity if MS want vista to look and work the way they want to, it is going need to be based on a Monolith Kernel
~Alan
yea but linux everything is a service. you can make changes and restart the said service, /etc/init.d/foobar restart
and voila! windows reboots far too many times then necessary.
The strongest desktop os in the world just got better. Cheers to microsoft.
And for all of you panicking about reboots.. theres a thing called scheduling you might want to check into. I manage a couple of machines and Solaris we never reboot unless we need to change system wide settings IE) A recent issue where /etc/system had to change required a reboot and on windows(2003) I only need to reboot when installing new applications. Either way, unless your 24×7 your users won’t be crying over 4 minutes at 3:00 am when they can’t access the application.
Remember, its *not* the os, its the applications that count. No one was ever hired to recommend an operating system for a business, it was to make business applications available. A reboot here or there isn’t going to kill you. Rebooting in general has continued to significantly decline in both desktop and server windows operating systems as the years go by.
– Microsoft Fanboy
Redmond start your copiers has come to fruition
From the screen shots, I have to say MS has done a great job of lifting everything from Mac Os X installation- it looks even easier than Os X in some respects. The biggest difference was the Window about anti-spyware-highlighting the major concern with Windows in general.
Its going to be tough to top this by Apple.
Almost as good as Apple and possibly tramples on Linux based on software availability.
Competition is good. Lets see what Leopard can do.
Redmond start your copiers has come to fruition
From the screen shots, I have to say MS has done a great job of lifting everything from Mac Os X installation- it looks even easier than Os X in some respects. The biggest difference was the Window about anti-spyware-highlighting the major concern with Windows in general.
Its going to be tough to top this by Apple.
Almost as good as Apple and possibly tramples on Linux based on software availability.
Competition is good. Lets see what Leopard can do.
Finally, a comment on this thread that actually makes constructive criticism of vista instead of the usual “micosoft sux” and “welcome to the 20th century bullsh*it
Everyone who has used Windows knows that it wants to reboot far more often than probably any other operating system out there, to the point that it’s simply annoying.
Now MS is finally doing something about it, which is a good thing.
Some people took this opportunity to make fun of the richest company in the world, so freaking what.
No need to get upset about anything, or write pages of comments attacking everyone who dares to point the simple fact out that Windows needs rebooting far to often, or spend hours of your life arguing this point with pretty stupid arguments.
Like many others have stated. Vistas continues to add features that have been available for Mac OSX users for quite some time now. It is refreshing to see Microsoft has still not lost it’s gift to copy instead of innovate. Maybe MS should just pay Apple et al to become their r&d department.
[/i]Maybe MS should just pay Apple et al to become their r&d department.[/i]
Why? They get that for free the way it already is.
Maybe MS should just pay Apple et al to become their r&d department.
Maybe the reverse should happen considering all the functionality other OSes (including OS X) have gleaned from MS.
install/update chipset drivers=Reboot
Install/update graphics drivers=Reboot
Install service pack=Reboot
Install/update directx=Reboot
Windows updates=Reboot
Install/update sound drives=Reboot (with my Audigy I had to)
People reboot Linux because they THINK it’s the same, I see it all the time. To me thats brainwashing.
MS has spent millions/billions to clone Os X look and feel into Windows. What a waste of resources!
Now that Os X will be able to run natively on Intel, may be MS should spend its energy to come up with something truly innovative and leave the Os end to Apple or Linux.
In a few years all Oses will look the same and most documents will have an open format (lets hope!). Then each person can use his Os of choice for computing
The round Windows logo button I like.
…are not only caused by it’s way to handle locks, it also has stability issues.
Just warn me when a Microsoft OS can do this :
# uptime
9:52am up 1207 day(s), 21:18, 1 user, load average: 0.10, 0.26, 0.32
(I know, this server is not up to date, but if it ain’t broken…)
Looks like MS is catching up, run forest run. Hehe
“Microsoft is working on a significant new feature for Windows Vista, known as Restart Manager, which is designed to update parts of the operating system or applications without having to reboot the entire machine.”
Ooooooooooooooooh, Ahhhhhhhhhhhhhhhhhhhhhhhhhhh, what will they think of next?
Well, I don’t know, but I’ve been told the streets in heaven are lined with gold. I ask you how things could get much worse if the Russians happen to get up there first; Wowee! pretty scary!
First to OSNews staff… please clean up in this friggin mess of people that obviously didn’t bother to read the article.
The main thing with the Restart Manager is that it can save the state of applications when it needs to reboot, and bring them all back to the state they were in before the reboot. This may or may not be possible if the application you are using is the one being updated (for obvious reasons), but for bug and security updates it means you don’t have to save your open documents, all your open browser windows etc. before rebooting.
If anyone can truly prove that other desktop operating systems can do the same as Restart Manager, please go ahead and do it.
I haven’t seen that in GNU/Linux, *BSD, OS X, or other OS’s that are common among normal people, so please prove that this is not new in this area (mainframes and the like not counted).
The OS/2 WorkPlace Shell could save and restore the OS/2 desktop in a similar manner when doing a Dual Boot to DOS. In fact, didn’t Windows 95 do a similar thing when booting to DOS mode?
“The OS/2 WorkPlace Shell could save and restore the OS/2 desktop in a similar manner when doing a Dual Boot to DOS. In fact, didn’t Windows 95 do a similar thing when booting to DOS mode?”
Actually OS/2 Workplace Shell and GeoWorks could do this. (I don’t recall any such behavior in 95.)
What was actually happening was the *Applications* were written in such a way that they could be given the ‘”Quit” command and they would save their own working document in a temporary document file. When the application was restarted they would restart with the saved work.
OS/2 and GeoWorks were built with the default that all running applications were restarted when the OS desktop came back up.
Ergo.. All applications came back up when the OS was booted into. Lovely.
Unfortunately, noncompliant applications (under OS/2) such as OS/2 command line (windows), DOS command line (windows), and Windows sessions were in effect process killed as they could not handle the quit command. They were restarted when the OS was booted into, but whatever programs or applications that were running did not have their context renewed because nothing was actually saved.
I expect the the Restart Manager under Vista will work the same way as OS/2 but older, non-quick-save enabled applications will simply be killed as in OS/2 but I at least expect Microsoft to recognize this condition and warn you about the applications which need to be manually closed first before preemptively restarting the system.
I don’t believe that even Microsoft can magically “save the state” of a running application if the application in question is going to be replaced as Windows is not an object oriented system (see Smalltalk).
Sounds like Save Session in GNOME to me, which saves the state of open applications.
“If anyone can truly prove that other desktop operating systems can do the same as Restart Manager, please go ahead and do it.
I haven’t seen that in GNU/Linux, *BSD, OS X, or other OS’s that are common among normal people, so please prove that this is not new in this area (mainframes and the like not counted).”
DragonFly has most of this functionality (it’s needed in order to be able to migrate processes between machines in a cluster):
http://www.dragonflybsd.org/status/report-2003.cgi
Under “Checkpointing”
Sure, DragonFly’s implementation isn’t yet finished, but hey, Vista is still in beta, so I think it’s fair to point out the fact that DF has this ability as well.
EDIT: I would also like to add that DragonFly’s implementation was “nearly complete” and quite functional back in 2003.
Edited 2005-12-02 18:32
At osnews anymore.
You reboot or you don’t, who cares? What’s the big deal?
I wonder if this will work anything like the way Crossover Office works with reboots. I use native Linux software for most of my needs, but when I use that occasional odd program that is only available for Windows, CXO generally lets me avoid booting into Windows. Just about everything asks for a reboot, and since CXO can reboot very quickly without touching my core system I barely notice the very minor inconvenience of a possible window or two being closed.
If MS manages a reboot process that is as good or better (don’t reboot the kernel unless necessary), and I believe they can now that they’ve decided to, it will be a very nice improvement to Vista.
Overall I think that even though there doesn’t seem to be one thing to make Vista absolutely necessary, there are enough improvements that all other things being equal Vista will be a nice upgrade from XP. The experience as a whole of using the OS should make Vista much more desirable than XP.
”Install updates without a reboot”
QNX has had this “feature” for along time, (beacuse it’s a well designd micro kernel OS)
Now people are saying that Linux has had this feature for a long time??
Linux is an old monolotic style kernel..
And one of the big drawbacks with a monolotic kernel is that a minor update to the kernel requires a reboot..
Micro kernels are very modular by design. (but ofcourse “core” updates requires a reboot of every OS)
Yes maybe Windows is playing catch with other modern OS (like BEOS, QNX) but not surley with old style Linux kernel.
Nobody whould design a new OS with a monolotic kernel today.
Now with new cpu:s coming from Intel and AMD 100% multithreading design is even more important. (with 100% I mean the whole OS )
I guess people really like to go back and forth on this one.
Some have said that it’s not Microsoft’s fault if every programmer out there decides that you should reboot after the most minor change to a system. I don’t completely disagree with this, but a feature isn’t very good if it isn’t used/usable.
When MS made XP capable of handling more system changes without a reboot than Win9x, they could have also made a change to the API so that Windows decided when a reboot was necessary instead of placing the onus on the programmer. Taking a different route, they could have changed the API/installer in a small way that forced programmers’ behavior to change by making a reboot less automatic. Either way MS could have made things simple enough that unnecessary rebooting was kept to a minimum with little hassle to the programmer.
Prior MS policy had gotten programmers into a habit of rebooting everything. MS could have helped them change their habits, or they could have supplied a tool that improved system behavior regardless of bad habits. MS didn’t do this. So, even though Windows XP is more capable of handling changes without a reboot, reboots are still far too common.
I run Debian unstable on a notebook. I run indefinitely (months or longer) without reboots AND YET INSTALL NEW SOFTWARE TO AND UPDATE the system EVERY NIGHT through Synaptic which not only updates system level stuff but applications, desktop environment, etc. The only time a reboot is needed is for a kernel update.
As for restoring a desktop after a reboot, Gnome (and I am sure KDE) have always provided for saving the desktop setup on logout and restoring it on log-in. That just goes with a multi-user operating system.
With SP1 for Server 2003, Microsoft introduced “hotpatching”, which is (from what I can understand) a way to patch a running system without having to reboot, due to a complex mechanism of state transfers. If you need to patch a few DLLs that are active, the hotpatching mechanism freezes them at some point in their execution, updates them, then transfers the old state to the new DLL, and continues execution.
As far as I understand, the only patching that requires a reboot now in S2K3 SP1 is that of the kernel. Vista is obviously taking this a great step further.
So there you have it, Linux fanboys … welcome to the 20th century. When will you be able to provide something better? If not for a while, then do *all* of us a great favour, and PLEASE *shut the fsck up*.
I love Opensource software and that’s have a big future, but today it’s just to complicated/too bad designed..(linux/x-window)
And look at this picture that explains Open Office dependencies…. http://www.yellowtab.com/images/news/ooo/dependencies.jpg and http://www.yellowtab.com/news/article.php?id=175
I remeber Windows dependencies hell with old windows, but with Win 2000 / Win XP I never had problem with “DLL Hell”
I really hope that the Linux community clean up the code/design right now! beacuse it’s just get harder to do it if you wait even longer..
Now that Os X will be able to run natively on Intel
So, you still need to buy the sys. from Apple.
Thanks for the examples of functionality similar to Restart Manager, in other systems.
Far from the barrage of implementations that some people want to claim ;p
Upgrade scripts??
Try to do that if the update regard some updates in the Linux kernal.. You don’t know what you are talking about if you not have tryed alternatives like QNX when it comes to modular OS designs..
“Nobody whould design a new OS with a monolotic kernel today.”
Really? Apple just changed from the Mach micro kernel to the FreeBSD monolithic kernel in Tiger and the speed improvement is noticeable. Microkernels are nice in theory, but generally seem to lag behind in practice. However, QNX seems to do quite well.
The NT and Linux kernels are both hybrids. Although, the Linux kernel may generally lean towards Mono and NT kernel towards Micro, but they are hybrids.
Where do you get that kind of info from?
Mac OS X still uses XNU – the mach micro kernel.
“The NT and Linux kernels are both hybrids.”
I don’t know where you’ve got that idea from, but Linux most definately is not a hybrid. It is a monolithic kernel, pure and simple.
It exists in a single address space (as do kernel modules when they are loaded), it uses message passing for absolutely nothing internally, and it does not have “well defined interfaces” (the Linux KPIs change more rapidly than in any other kernel I’m aware of).
It represents the pinnacle of monolithic kernel design, and anyone who claims otherwise is either lying or completely clueless on the topic.
“Really? Apple just changed from the Mach micro kernel to the FreeBSD monolithic kernel in Tiger and the speed improvement is noticeable. ”
Uhm, no. XNU (the Mac OS X/Darwin) kernel is (and has been since before its first public release) a hybrid*; that is to say that it is not a pure monolithic kernel, nor a pure microkernel, but a mixture of the two concepts. It is also very much still based on Mach. They’ve just updated the BSD layer they’ve welded in to code derived from FreeBSD 5.x because of the fine grained locking work that’s gone into FreeBSD.
It is excedingly unlikely that Apple is ever going to drop Mach as the core of their kernel, and just as unlikely that they will move the BSD portion of the kernel back into userspace. Mach provides more fucntionality that they rely on and for them does a better job of it than any of the alternatives (message-passing, kernel threads etc.)
* It is a hybrid as it is based on a microkernel (Mach) and uses the interfaces, and especially the message-passing internally for quite a large number of things, IPC in particular, and for things like their XGrid application. It also contains more functionality than would a pure microkernel for example many drivers, the BSD layer, network protocols etc. are in the kernel, wheras in a pure microkernel, they would not be.
Hybrid kernels are the here and now (and likely the future) of operating system design, with only QNX, GNU, Linux, and most of the BSDs being the odd ones out.
Edited 2005-12-03 10:32
There is no such thing as “hybrid kernel”, XNU is pure monolithic by design having drivers + BSD + Mach component in kernel-mode. “Hybrid microkernel” is just marketing propaganda from Apple and MS. Having Mach in XNU has nothing to do with it being more microkernelish or not since THE ONLY CRITERIA to distinguish microkernel from monolithic kernel is the amount of code running in kernel-mode (ring0)
http://www.kernelthread.com/mac/osx/arch_xnu.html
“XNU’s Mach component is based on Mach 3.0, although it’s not used as a microkernel. The BSD subsystem is part of the kernel and so are various other subsystems that are typically implemented as user-space servers in microkernel systems.”
Another common misconception is that NT kernel has some “microkernel component” inside, which is a complete nonsense, it’s even more monolithic than traditional UNIXen running GUI (GDI/USER) in ring0.
I use both Windows and Linux and on the Linux side, I’ve never required a reboot after installing updates. Never. I have been prompted to install a new kernel on one of my Ubuntu boxes and it doesn’t require a reboot unless I want to use some feature of the new kernel which I probably don’t since I went from 2.6.12-9 to 2.6.12-10. I haven’t rebooted yet.
Why do you people constantly bother posting on a Vista thread if all you have to say is just windows bashing? I’m sick and tired of feeling like I have to be validated by the (*)nix community by making up some stupid excuse just to avoid being ridiculed. Sure MS has made some stupid mistakes in the past, but even now that they’re making strides towards rectifying those mistakes you people still won’t give us a freakin break!
Oh, like Linux hasn’t had any problems and missteps and wrong moves in its development. Oh Linux is immortal I forgot, God uses Linux for his print servers!
Come on, could we please just stop bashing each other’s choice for OS and focus on the reason we’re all here, love for OS technology? No matter how long it’s taken, we should ALL be excited to see another OS become more full featured, and therefore competitive, ESPECIALLY a Microsoft OS! Isn’t that what most ppl whine about? They want competition? Well here goes, MS is making a move to get up to speed to compete feature-wise for once instead of just thru marketing and strongarm tactics.
I think that scares FOSS ppl…
Can’t we all just get along!
Glad to see this feature added. It’s been needed for a long time. The only OS I have ever used that didn’t require frequent restarts for maintenance and installation purposes was BeOS. If a driver crashed on you, the server for that driver would restart, leaving the rest of the system running erstwhile.
This feature doesn’t mean you no longer need to reboot.
It just means Windows Vista won’t need to reboot when it doesn’t need to. This is no different from BSD, Linux, Solaris, etc, etc.
Is it a new feature?
For Windows = Yes
For everyone else = No
Features like this have been experienced by users of other OSs for many years, its just taken MS so long to to actually include something similar…That they think its a new thing.
Everytime when MS announces something new, it seems to imply they lock away their “engineers”, developers and such from the world. When they “develop” something, they think its new…Its more than likely to be already done by everyone else!
As for Microsoft touting about “new” features? Well, I believe it when I experience it and see what are the limitations in their implementation. Because knowing Microsoft, there’s always a catch somewhere. (I don’t trust their marketing or their public announcements).
looks like windows is moving forward
checkout the start menu it is so nice, the bit i couldnt take a screenshot of is the ‘windows tab’
third icon from the left
you’ll all get to see it soon enough
http://anyweb.kicks-ass.net/computers/os/windows/longhorn/lh5259/po…
cheers
anyweb
It’s about time, this functionality has been availabe using a piece of free software for several years. Can’t remember where I saw it, but I’ll keep looking.
That is a wonderful new technology. That is defintely a show of that Microsoft ingenuity. This is just incredable. I cant wait to see what else they are able to come up with. Maybe next they will eliminate the need for a crash every time I open 5 or more apps. That would be heaven.
If a program crashes with a file open then you can no longer access that file. On Linux this does not occur.
Spyware on XP can stop itself from being deleted by locking the file. That can be overcome but in Linux that would never happen, superuser has the power to delete any file.
It is very good to see Microsoft fixing these problems, in the installer stage. Hopefully they will go further and fix the examples I have listed above. Does anyone know if these problems exist in Windows 2003?
afaik, there really isnt much of a way you could possibly have a system that could update the kernel without rebooting, at least without causing instability. im fine with rebooting for kernel, its the only thing ive ever had to reboot for. its good to see windows is catching up on that though, you would think with the outrageous amount of funded developers for windows this woulda made it here in 9x days, but cant knock em for trying to make things right
, dont knock linux/unix, or its users for having something simular first, if it werent for *nix competetion, microsoft would see no reason to offer anything new, or useful, they are a maintenance company, keep the same old stuff until something threatens its appeal, then throw millions at it. I dont hate microsoft, but they need a swift kick in the ass from time to time. And this is proof it works.
Microsoft is treating the effect, not the cause.
Windows reminds me of a Rube Goldberg device & this
so-called improvement is akin to adding a balloon tied
to a rabbit’s ear, that when the balloon is released, the
rabbit kicks over a tea kettle, etc.
The only time a reboot should be necessary is when the
BIOS ROMs are replaced (or some other essential hardware!)
In my experience I never have to reboot Linux. I put my home and work PCs through their paces, adding and removing software willy nilly, and nothing requires a reboot, ever. The only time I see the power-up screen is if there’s a power outage. I do anticipate I’ll need to update the kernel at *some point*, but what’s one required reboot in a couple of years?
My experience here contrasts dramatically with that of my time with Windows.
As for saving your software state on a reboot, logout or restart of X Server, KDE does a pretty good job of it. (I have no experience with Gnome)
DOWNLOAD Windows Vista build 5259: http://windows.czweb.org/show_article.php?id_article=379
I’m really amused with the enormous amount of bullshit that penguins seem to direct towards Restart Manager. Have you EVEN READ WHAT IT DOES? No, and if you did, you’d new that it has NOTHING to do with kexec and this hasn’t been part of any OS so far.
For example, applications implementing MSI 4.0 features like MsiRMFilesInUse/RegisterApplicationRestart could gracefully be stopped, their state could be saved and restored while the update of core system components they are using is beeing taken care of.
That meens that if youre burning DVD with Nero while installing OS updates, it will be paused for several seconds, restart it’s main EXE component and continue it’s work WITHOUT INTERRUPTION. Same goes for ANY app, and this is NOT just “reopening files that were opened in Word”.
As for the NTOSKRNL and rebooting, tell me if linux has something like hot patching? Ie: patching a kernel in-memory requiring no reboot at all, by detouring 2 bytes before prologue of every function?
Of course it hasn’t, since that it hasn’t been able for 15 years define a stable set of kernel ABI. Windows has it, Solaris has it, all well-engineered OSes have it..but linux doesn’t.
Just to mention, hot-patching is integrated into MSI 4.0/Vista OS updates. To apply patch in linux you MUST restart, either via kexec or recompile + restart manually, for Vista, patching NTOSKRNL, drivers or ony user-mode DLLs is just a matter of seconds and done via syscall NtSetSystemInformation/SystemApplyHotPatch.
Just to mention hotpatchig is already present in XP SP2/WS2K3 SP1, but NOT utilized at all, at least not as much as it should.
For more information on MSI 4.0 and RM read:
http://microsoft.sitestream.com/PDC05/FUN/FUN222_files/Botto_files/…
Sorry, I was reading the first three or four pages of comments and zealots of every possible colour are defending LInux, *BSDs, Windows, etc… the fact that Windows has a new service to control things that need to reboot is the point. And is most welcomed. No need to compare completely different OSes.
I have linux Servers. Mac OS on my POwerbook and Windows on my Main .NET devel box/Gaming box.
Now try to play a game on linux (yeah… sure).
Best tool for the job guys. Stop crying about linux being able to restart a driver. That’s possible thanks to a Modular Kernel. Period. But who really cares? In the end, weather you whine or not, Windows Vista users, will be more pleased in overall experience; time will only tell how true this “less reboots” thing is under Windows.
Now I wonder, if Windows had this feature, were completely themable, etc.etc.etc., what Linux users would complain about?
Linux is lovely, in command line, with no X installed, serving daemons. When using X with some DE/WM, it’s a huge beast of unconnected -sometimes buggy- applications and stuff you don’t know or care about. Unless, of course, you’re just a droid who needs to use OpenOffice (or something) + read your email…
More than that and you will have to trim and customize the monster (and you will be able to) to suit your needs (because you can). But that doesn’t change the “unconnected” beast fact.
Me? I am not upgrading to vista, why would I want to. XP works.
On Windows XP, I noticed that Windows Update sometime requires a complete shutdown of the OS. Is it for the core? Hopefully it won’t be the case for Microsoft Windows Vista.
The reason Windows reboots, as far as I can see, is that it reduces their testing matrix for patches, which is already enormous as it is.
What’s the point of delaying potentially important patches to fix security issues just so that you can test that they interact properly with in-flight requests from already running applications. The engineering cost of doing this is far greater than the slight benefit from users not having to reboot as much. Enterprise users mostly install Hot Patches and have clustered servers so that reboots are unimportant.
My prediction is that this feature will not be fully implemented in Vista. The furthest they will go is to allow apps to support saving and restoring state upon reboot. This will probably be an opt-in mechanism, so it won’t restore an app that doesn’t support it. Maybe they will be able to piggyback this onto their hybernation support, but I think that’s really complicated and pointless, and I think it’s impossible to transfer state from in-flight requests from the old code to the patched code.
Really, who besides you cares how long your linux box has been running? If you want a system to be solid, reboot it after patches to ensure that all software picks up the patched library.