Windows 7 will not go gentle into that good night: Ageing OS refuses to shut down It's not only end of support that Windows 7 diehards have to contend with. Late last week a new problem emerged – systems that refuse to shut down. Complaints have been widespread on Reddit, Microsoft's official Answers forum and on on SevenForums. Some users also reported other issues, such as not being able to view their …
It was fashionable to pretend 25 years ago that Windows was a "real OS" that, like mainframes of the day, needed to be properly shut down if seriously bad things were not to happen to users PCs. However, many users didn't get the message and continued to turn the thing off with the power button. And even those who didn't had to contend with folks elsewhere in the building doing stuff that tripped their circuit breakers and with power failures caused by lightning or drunken drivers smacking utility poles or other slings and arrows of outrageous fortune. So Windows remained pretty much impervious to unscheduled power off. Unix handles it pretty well also BTW.
My recent experience with Windows is limited, but I'm pretty sure that pressing the power button starts a shutdown (or maybe a suspend to RAM, which is what my Linux boxes do).
Unless you hold the button down until the firmware does a hard power-off.
This can be set in the "Power Options" control panel.
And I'm not so sure about altering a setting via Group Policy. Unless someone can present evidence that this specific setting is being altered by whatever 1st/3rd party component is causing the problem, it might do more harm than good. AFAIK this is equivalent to disabling UAC which causes all processes to run at the highest integrity level. Of course, you weren't running with an Admin account in the first place, were you?
I just think that if there's half-a-dozen other ways to get the computer to turn off peacefully, taking a hatchet to an important security restriction seems the least advisable one. UAC has saved my from my own clickhappiness at least once.
AFAIK this is equivalent to disabling UAC
The article says to set Run all administrators in Admin Approval Mode to Enabled, which enables UAC. If it's not already set to Enabled, then the system has UAC disabled.
FWIW, my Win7 machine has maximum UAC (all UAC GP settings set to the most-secure option, prompting for credentials on the secure desktop1), and I haven't seen this issue.
1Which, yes, is still far from bulletproof; but it does increase the attack work factor.
"[...] but I'm pretty sure that pressing the power button starts a shutdown [...]"
The default actions of the power button or lid closure on Windows on laptops is often not particularly sensible. Often everything is set to "sleep". My strategy is to make sure that the power-off button initiates a tidy shut-down - which will often get you out of a stalemate situation without a crash-down.
Later versions of Windows have hidden the explicit "hibernate" function. The composite "sleep" followed by an implicit later "hibernate" seems like it would power up to do the latter - when the laptop is in your bag.
W7 introduced a blank screen while the hibernate process was running. Some laptops have minimal disk activity LEDs - and there is a risk of the user subjecting it to rough handling while the disk is still busy writing.
The default behaviour for a fresh install of Win7 (and Vista/8/10) is that the power button does a shutdown. I've seen pre-installed machines which have it set weirdly, but that's down to the OEM.
In Win 10, the option labelled "Shutdown" is in fact 'hibernate' in order to speed up startup times. This leads to Windows reporting an uptime that is much longer than the time since last power on.
(You can force a 'real' shutdown by holding shift before you click 'Shutdown', or use the command line shutdown /s /f).
I really dislike that "shutdown isn't shutdown" behaviour in Windows 8/10. In no small part because of some people who when I say "restart" instead go and use the shutdown option then manually power back on.
Thankfully it's easily disabled through the Power settings. (GPO/Regkey also available, obviously.)
I really dislike that "shutdown isn't shutdown" behaviour in Windows 8/10. In no small part because of some people who when I say "restart" instead go and use the shutdown option then manually power back on.
For me the issue came from the point of doing data recovery work, or backing stuff up before working on the disk. Often the easiest way is just to boot from a Linux USB/PXE etc and copy over what you need, but when Windows is hibernated most distro's wouldn't mount the disk even as RO (should've played with that more and set it so the general user gets RO but a root user gets RW but I digress).
Some BSODs could be fixed by a Linux-based registry editor. The W10 startup menu wasn't very functional unless you'd told the machine to go into "safe mode" at the last shutdown before the error occurred (that was one of the stupidest things MS did, requiring safe mode to be initiated from a perfectly working desktop rather than from any attempt to start the machine).
I'm glad to be all-but right out of the industry, and I do not do support on W10.
I'm no fan of windows on server but oh you sweet summer child, Windows NT was pretty solid as long as you knew not to install crap on it. Linux/Unix/BSD was better to be fair but lets not try to change history.
p.s. Fuck you Solaris, I never liked you.
"[...] needed to be properly shut down if seriously bad things were not to happen to users PCs."
The introduction of W95 soon taught us that the previous practice of just powering off was not a good idea. Disk file store corruptions were a pain.
A friend recently had a problem where her W8.1 laptop's battery wasn't getting charged properly. This caused untidy automatic power-offs every time she tried to use it. I fitted a new power socket - but the hard disk file store obviously had a subtle corruption that stopped many applications from running. Repriming from a system image archive was the only solution.
Back in the 1960s our mainframe OS regularly had a problem of file store corruption if it was crashed down before the disk copy of the sector usage was synchronised. Eventually this was circumvented with a solution that merely "lost" some free file space - by marking sectors as "allocated" on the disk before they had actually been used.
An operator command did a scan and recovery of the lost sectors. MS OS have "chkdsk" which appears to be the solution to a similar potential problem of synchronising sector allocation.
Windows got a lot better at dealing with an unplanned shutdown when they shifted to NTFS as the default file system. FAT doesn't cope with power loss nearly as well, and EXT2 had the exact same problems.
Of course, there's still a slight difference between a hard shutdown (eg, by holding down the power button on a modern machine), and a complete power loss, because at least with a shutdown, there's still power to the harddrives which prevents problems with a head crash.
Modern drives shouldn't suffer head crashes from sudden power losses, unless they're already having other problems (e.g. failing drive electronics). It only takes a tiny bit of power to park the heads, and it's easy to supply that with a capacitor on the drive controller, along with the logic to self-park.
I have a Dell Latitude laptop which hasn't been able to charge its battery for years, due to Dell's rubbish engineering. It's suffered numerous sudden power failures, since it's dependent on wall-socket power. It's never had a head crash. Anecdotal, I know; but it shows that power-failure head crashes aren't nearly as common as they were in the 1980s.
Modern drives shouldn't suffer head crashes from sudden power losses, unless they're already having other problems (e.g. failing drive electronics). It only takes a tiny bit of power to park the heads, and it's easy to supply that with a capacitor on the drive controller, along with the logic to self-park.
It's actually less technical than that even... The head arm is lightly spring-loaded, sufficient to withdraw it from the platters and onto the parking block in the event of a power loss. Still, I have seen a number of "click of death" drives that had to be opened to get them working again (easy to build yourself a small "clean box" for that work - look at a sandblasting booth and get your inspiration from there (only instead of feeding sand+air in your use a vacuum cleaner to suck it out, and a set of dust filters on the other side so only clean stuff comes in).
TYVM for the Dell charger link BTW. Has annoyed me no end how it refuses to charge the battery even when the laptop is turned off, even if you have a laptop that wants a 65w PSU but you only gave it a 60W, or wants a 95 but you only have a 65. Bloody Dell - you can still charge the battery when the thing is turned off even if the charger only gives 2w! I did see a few Dell's where the power socket suffered a broken solder joint - often the power pins were fine but the charger ID pint was the one out of commission. Did at least lead to extra work for us - the hardware repair of the socket plus the repair of the corrupted disk where Dell could've handled that more gracefully!
So if the OS is past end of life and is no longer receiving patches or updates, what CHANGED to make the shut down no longer work? Like, literally, what CHANGED? what update was pushed, what bytes were altered, what logic and code was changed to make it behave that way? Operating systems don't just rot or break by themselves ffs.
So if the OS is past end of life and is no longer receiving patches or updates, what CHANGED to make the shut down no longer work?
Exactly. The line from the article "[...] see increasing numbers of mysterious issues with Windows 7 as it becomes less well supported [...] seems completely wrong to me...I would have expected to see no change in behaviour as no updates are being provided for the OS. If nothing changes, then nothing changes
That's true if the software on the OS does not change. If there is stuff like the Adobe service that automatically updates though, or you install new applications, that is when the problems occur.
Thing is.. Much software still supported XP for years after it went EOL, and I believe there's more differences between XP & 7 there there is between 7 & 10.
This is not even a month after EOL for 7.. I honestly don't believe that there would be that much of a change in these systems...
And why is Adobe changing stuff that affects these keys (or is fixed by changing these keys)? Has Adobe actually released any updates in the last few days (if memory serves me, almost certainly yes, several gigabytes, each time, each hour, or so it seemed - been years since I've had anything more than "Elements" on a machine!)
It does not feel right that updates would be released this early that 'accidentally" cause this issue, although it is a given that both MS and Adobe have had lets just say "some rough patches" with their QC depts...
"Bit rot" affected cds and dvds as far as I recollect. Moisture permeating junction between the layers of plastic "sealing" the layer of organometallic recording medium allowed wee little organisms to actually caused the layer to decay. Well, essentially they ate the "organo-" bit. There used to be some nice microphotos of "bit rot" occurring and at least one time lapse video.
"Like, literally, what CHANGED?"
Sounds like Adobe doesn't do quite enough testing on older (less marketable) platforms and so their DRM service was updated containing a regression exhibiting shutdown issues on Windows 7.
Luckily, like us, most users of Adobe software are using cracks which also lobotomises most of the DRM service so they never actually run into this bug.
I suspect that Adobe didn't do any testing on Windows 7 (or 2000, or XP, or 3.1, etc.).
I also suspect that the same can be said for any number of applications that install services.
At some point people need to realize that non supported OSs need to be replaced or they need to be paired with non supported versions of applications.
And it's not just Windows; if you're running Linux with a 10 year old kernel, you probably want to upgrade that too.
"At some point people need to realize that non supported OSs need to be replaced or they need to be paired with non supported versions of applications."
It's a bit rich to expect them not to be testing on an OS that only just went out of support though. After all, who's to say how long they've been working on the latest patch and when they originally planned to release it. Especially when they KNOW they will pushing patches out to said OS. If they don't test against an OS, then their patch installer ought to be either NOT pulling in new patches or popping up a warning the user that the patch is untested. After all, it's them that do everything they can to switch auto-install back on every damn time an update gets installed.
And it's only out of support for ordinary retail. Microsoft will still be engineering patches, testing and deploying to POS, embedded and paying customers. Thus it's purely about money grabbing from big customers and forcing people to change 3 years early.
XP POS support didn't end long ago.
>And it's not just Windows; if you're running Linux with a 10 year old kernel, you probably want to upgrade that too.
It depends. Most of our problems with compters come from the assumption that a computer is a device with a screen and keyboard that allows a user to interact with the outside world using either a general purpose web browser or specialist browserlets that are known as 'apps'. If the computer's doing a job then if it worked 10, 15, 20 years or more ago it will still continue to work today.
Continuously upgrading makes software unrelaible because the upgrade cycle becomes shorter than the test cycle.
Operating systems, like all other software, absolutely do rot and break. Not by themselves, admittedly, but the whole point is they're not "by themselves".
Even after the OS has stopped getting updates, other software on it (e.g. Adobe crap) will keep getting updated. Or users or admins will change settings. Or someone will plug in a new mouse, and a new driver will be downloaded. Or whatever.
There is no realistic way of putting a freeze on the whole system, so if a single part of it is frozen, it will gradually, but inevitably, grow more and more unfit for purpose as it gets left behind.
Have you not noticed over the decades, that for every version of Windows, as its "uptime" increases, it eventually becomes slow, erratic, and generally unstable? In an anthropomorphic way, I sometime say it's getting "tired", "cranky", or "unhappy".
When that happens, it's time to reboot. If you wait too long, Windows hangs or blue-screens on shutdown, requiring a power off. After a reboot, Windows runs better.
Now, why do you suppose this happens, this gradual decay in stability? If your assertion were true, if I didn’t add/remove hw or sw, Windows should run forever, with no decay. But that's not what happens.
I can't tell you why — I could speculate about buffer overruns and stacks being over- or under-pushed or popped, variable scope errors, etc. But, I’m just guessing.
Furthermore, unless hardware-enforced protection of executable portions of files is enabled for all software, your buffer overruns can permanently change software.
My windows 7 machine has been on for 193 days and is running fine.
We used to have thousands of windows servers running VBAK back in the day, and they would have uptimes of years and run fine.
I suppose it may be different for home PCs, where users are running many instances of different programs, increasing the chances that one of those is going to misbehave at some point, or a driver leaking memory etc.
This post has been deleted by its author
The point is that they can't upgrade the distribution because the newer distributions don't support their software. They are just as locked in as anybody else who is using old, unsupported software for business-critical applications, and as such are in the same boat as anybody else using proprietary software. Linux doesn't fix that.
Until they "discover" the budget to fix the problem, the systems will remain in production and they will rely on the archival backups to get them back up and running should the entire house of cards fall down around them. At least in this case the backups and recovery procedures have been tested, most companies in this situation can't even say that.
Have a look around.. You'll find many pros using Linux for their day-day stuff. Film making, picture editing, publishing.. Been some interesting threads on here in recent weeks...
As to small.. The most used kernel in the world by far is "small"? Windows has the desktop, Linux has servers, phones, tablets etc.. You know, the stuff people use today..
And didn't "multimedia" die out in the mid 90s? :)
WIndows is not an OS, all versions of Windows are just a GUI. The last Microsoft operating system was MSDOS which, if you remember, they built from QDOS which was copied from CP/M.
MS-DOS was never an operating system worthy of the name.
All the NT-series versions of Windows contain an operating system -- Microsoft employed Dave Cutler, from dec, to go and write it for them and he had some experience in the art -- but that operating system has been borged with a GUI toolkit and a window manager, with a graphical desktop held on top with sticks and old chewing gum.
There's an operating system in there but you can be excused for missing it.
Was Win NT 3.1 the first version in 1993 because after IBM & MS ended the OS/2 partner ship that MS sold MS O/2 with bundled MS Lan Manager from 1989 or 1990 onwards?
NT 3.1, 3.5, then 3.51 (patched to have fake Win32 APIs added so Office 95 wouldn't run on Win 3.11 with Win32s extensions)
NT 4.0
Win2000, Win XP, Win 2003, WinXP for Itanium etc all NT 5.x
Vista and Win 7 are NT 6.x
Win8x is really NT 7.x
I'm not sure if Win 10 should really be NT 7.4 or NT 8, they certainly had to skip 9 because of stupid programmers in the Win95, Win98 days checking the wrong string for 9. Win95 was basically Windows 3.1 on DOS with all the 32 bit drivers, Win32s, an improved VFW, an API to allow porting DOS games (DirectX) and a new GUI shell with less good Explorer replacing Filemanager. Not a real OS. No NT based security, no NTFS, no HAL, no NTVDM (16bit Windows & DOS in a window ran native, but NT used a Virtual machine and WOW API translation, as on Alpha). Also Win3.x / Win9.x / WinME only ran on x86 mixed 16/32, so was poor on Pentium Pro (really a CPU intended for 32bit only), NT ran on Power PC, Alpha, Alpha 64, MIPS, x86 in 32 bits only (no Switching to 16 bit). NT (1st XP pro 64 version) ran briefly on Itanium, this was EOL before XP was. There was also x86-64 AMD version of XP pro later, maybe similar to Server 2003.
It's been all downhill since 2003.
This is Microsoft we're talking about - the Comcast of software pushers. Just yesterday I was looking for the real meaning of one of their typically unhelpful error messages. Several MS employees and graybeard-equivalent experts on the Microsoft forums would lead the seeker-of-truth down many twisty passages which didn't help. A third-party expert-ish site identified the true cause, but claimed the setting that corrected it was tricky to get to. A brief comment to that article, by someone with no letters after their name, showed the three-click method of finding and fixing that troublesome setting.
By removing the useless Microsoft shills from the pool, I'd claim that Windows 7 is now better supported than a month ago.
Several MS employees and graybeard-equivalent experts on the Microsoft forums would lead the seeker-of-truth down many twisty passages which didn't help.
I only know of "Nethack" (IIRC that's what it's called) by reputation only. I have never played it or even seen it.
However, I feel that I could ace it after a few experiences with MS "support".
If any one is capable of taking the most convoluted route from a simple question to a ridiculously stupid answer they get hired by MS support and immediately promoted to senior support levels.
Several MS employees and graybeard-equivalent experts on the Microsoft forums would lead the seeker-of-truth down many twisty passages which didn't help.
http://www.tensionnot.com/jokes/customer_service_jokes/microsoft_tech_support
(although my favourite is still http://www.rantnroll.com/html/gates.html#divine)
Something I never understood: ESR officially ends in 2023.
Yet officially its being supported "as long as is needed" but not clear what happens when that also expires.
Are we going to have a major disaster when mission critical systems just stop working or are M$ going to simply disable Internet support apart from essential services for W7 holdouts?
This actually happened to me twice: laptop *1 died completely and won't even see the mPCIe card after an update: laptop *2 is offline for now to stop it getting any more evil updates (tm) which was actually in hindsight a wise choice as it was already having strange problems like partial shutdown and trouble coming out of Sleep evidently due to its 51.4% remaining battery life.
Alas *1 was the machine I used all the time for my SDR because it has a good battery and reasonable VT-64x support along with 4GB of Crucial 1.5V RAM.
>> What if I have to replace a hard drive?
This was exactly my concern, which is why I'm running my now unsupported copy of Windows 7 in a totally portable virtual machine.
And the VM is backed up too. So if the working copy catches a cold, I can just revert to the backup.
It happened with NT, it happened with 2K, it happened with XP, it happened with 7. After a few years, a Windows PC seems to acquire some sort of crud that gums up the shutdown process so that you have to wait longer and longer periods of time to shut one down.
And people wonder why I avoid Windows like the plague...
In 20 years I never had a Win OS hanging on shutdown. That's usually what my linux boxes do since 5/6 years tho, before I start having arguments with SystemD.
My win boxes hang on startup usually, but 1) a good reinstall cycle usually fixes that for a year or so and 2) not happening anymore since 10 (well, happened once but it was the infamous borked update, rollback fixed that).
Something is fishy here with the timing. I am not sure who has been 'updating' their code after the last patches, I do not support any Windows boxes right now. The usual suspects for crappy updates have been fingered. But what is missing is more details of what was going on and it is quite possible the user might be unaware of what exactly happened.
This goes back to Windows For Workgroups, a.k.a. Win 3.1.1.
A time bomb installed in the software killed its network functionality in late 1995, some time after the intended release date of Win95, but before the actual (delayed) release date of Win95. Tweaking MemMaker disabled it and allowed the (crude) network stack in Win 3.1.1 to start working again.
How do I know? I found it myself, and saw that it included a workaround (a MemMaker change), which I then deployed for several customers whose networks had mysteriously all failed on the same day. Others found pretty much the same thing, confirmation that we weren't having a bad dream.
My (now air-gapped) Win7 dev box hasn't had this problem. I wonder if not installing last month's patches has anything to do with that?
My (now air-gapped) Win7 dev box hasn't had this problem. I wonder if not installing last month's patches has anything to do with that?
If MS were doing this competently, they'd've set it up some time back, not on the last update.
And if MS were doing this really competently, not all machines would fail at the same time either, and perhaps not all in the same way.
Thankfully MS aren't that competent these days so probably actually a true bug... :)
Not only Nexus...
I had a Motorola Razr M which I loved. After an update (4.2 to 4.4?) battery life literally dropped by half. It only took a minute of searching to find this was a near universal problem.
Neither the first nor last time this has happened. Something New!Shiny! comes out and older models mysteriously start developing problems.
"The favourite solution is to tweak the UAC (User Account Control) settings with the Group Policy setting "Run all administrators in Admin Approval Mode" or the equivalent registry setting."
No, not a good idea if the user is an administrator. That would be like running everything in a root shell, only worse, you would be running Xorg and all GUI programs as root as well.
Am I the only one receiving Win7 updates on a daily basis since 14 Jan? Security Intelligence Update for Microsoft Security Essentials - KB2310138, running from Version 1.307.2308.0 (14 Jan) to 1.309.742.0 (11Feb). I also received four other updates on 16 Jan - KB4534310, KB4535102, KB4536952, KB890830.
What's going on?
A not-so-recent interview with GoG highlighted that the hardest era to emulate is the late 90's / early 2000's Win98-(insert 3d-accelerator of choice) combo. There is a whole bunch of software from that time that won't reliably work on other platforms without much more capable emulation.
I'm not worried about office apps or migrating data to current Office platforms; because VM's are very well developed and supported for Win7.
There are, however, tonnes of embedded systems using the "wrong" OS that need attention. They certainly won't be being patched in the field, and aren't compatible with VM type solutions too.
When is support not support? When it makes things worse than before it was applied.
Why the need for continuous "support"? Because the product is so faulty that it would fall over without it? Or maybe because terminating support is the optimum way to ensure profitable churn, and if there wasn't "support" there'd be nothing to terminate. Maybe a mixture of both?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
