I worked in Antarctica last year, a "winter-over" contract as combination Network Engineer/Glacier Search and Rescue (GSAR) team member, at one of the very small U.S. research stations.
In the atmospheric/climatology lab there are a number of computers running very specific versions of Windows (down to specific patches) tailored to really obscure science research software packages, some industry standard, most custom-rolled and tailored to that project. Which means they're very fragile and need to be left alone.
Of course Microsoft blasts Win10 installer files down our very expensive, very slow satellite link, and proceeds to force Win7/8 systems to Win10 which broke a few multi-million dollar experiments in a place that takes MONTHS to get to in the winter. So effectively killed the research.
And yeah, you can argue "well you should've disabled Windows updates etc etc" but in REALITY, not perfect-theory world, you should be able to trust your OS manufacturer not to force-install a major OS revision without your consent. And in the old days of Microsoft, you could. Now they just say "f--- you, we're installing it regardless what you want."
> Of course Microsoft blasts Win10 installer files down our very expensive, very slow satellite link […]
The lesson there is that you should always quarantine computers that need to be pinned on very specific versions of an OS, especially in such a precarious setting! Those computers should not have been able to communicate over the internet at all, except for those cases where communication is part of their purpose (e.g., monitoring the experiments), and that should happen via a dedicated VPN allowing only the traffic necessary.
I'm surprised this isn't the standard operating procedure for this kind of case.
Could be it had been sitting down there for years without any incident.
Best i recall, MS was very very insistent with their Win10 "upgrade". To the point that they relabeled the installer patch as critical (aka, reserved for pushing 0-day patches or similar) after people found they could change certain settings to make it go away.
> Could be it had been sitting down there for years without any incident.
Well, sure, and if they ran a critical machine with no backups whatsoever it might have also been fine for years, but that hardly makes it reasonable practice.
Out of curiosity, would it not have been a better idea to build your experiments/software on Linux or a BSD variant? Was it a lack of software drivers for your equipment? Given the precise control you can get over the operating environment on these OSes, you can pretty much ensure that the environment your programs run under will never change.
They weren't "my" experiments - I was the station network engineer with double-duty as helpdesk/IT person.
Imagine this: you're a broke grad student and you're given $xx grant money to conduct a certain experiment in N months. What are you going to spend your VERY limited time+money on? Finding a good Linux hardware developer to roll some custom solution that no one can understand or support? Or just throw together some hacky Python or .net MVP that will run on Windows which practically anyone in the world can troubleshoot with basic knowledge?
Linux is great and all (runs on all my systems except one), but it is NOT the best option when there are heavy time constraints and the system is going somewhere very remote where there will be one, MAYBE two people with basic IT knowledge to troubleshoot.
What are you going to spend your VERY limited time+money on? Finding a good Linux hardware developer to roll some custom solution that no one can understand or support?
Just install some distro and "just throw together some hacky Python". If you had some hardware and measurement instruments that require Windows for drivers then I could understand it, otherwise it's just negligence.
> or .net MVP that will run on Windows which practically anyone in the world can troubleshoot with basic knowledge?
In my experience people who claim that they could troubleshoot Windows is usually just that - they claim it but rarely can they actually fix anything.
I don't know, maybe Windows is that much better and easier to use for everyone and I just can see it - who knows...
> If you had some hardware and measurement instruments that require Windows for drivers then I could understand it
It likely is. I have read similar stores about having to refurb an aging 386 or 486 running DOS. This because it is being used to operate a very expensive particle sensor, and the company that made it is long gone.
And if you want to replace said sensor you first have to get the funding to do so, and then you have to shut down the lab for a year to run calibrations so that the results from the new sensors can be compared to those of the old one.
EDIT: Didn't recognize the username at first. I still have some doubts, but I do remember you talking about going to Antarctica in the Tron subreddit. I recently used Tron, and I hope that there isn't some undocumented part of the script that is disabling automatic updates.
I have some doubts about this story... The forced Windows 10 update was only sent to non-enterprise versions of Windows. I highly doubt any of these experiments were running the home editions of Windows due to the lack of control. Also, I am pretty sure that running home versions of Windows in contexts like this is against the Windows TOS.
> I have some doubts about this story.... I highly doubt any of these experiments were running the home editions of Windows
I guess you've never worked with budget-constrained grant projects before.
> I do remember you talking about going to Antarctica in the Tron subreddit
7 months on ice.
> I hope that there isn't some undocumented part of the script that is disabling automatic updates.
Tron doesn't perform ANY undocumented actions. Unlike most the people defending Microsoft's actions in this thread, I am pro-user and seek to minimize negative impact to users in any project I'm involved in.
> Also, I am pretty sure that running home versions of Windows in contexts like this is against the Windows TOS.
If you're telling me there are people in the world using Windows in violation of its TOS I'm shocked, SHOCKED to say the least.
So, no offence, but this reads more like you and your colleagues destroyed a multi-million $ project by trying to save a few dollars by using Windows Home instead of Enterprise?
None taken, but you obviously didn't even read the post.
These project machines are shipped (or physically hand-carried) by grad students from random universities all over the world. They come from who-knows-what IT department running who-knows-what software packages, often cobbled together JUST enough to get it working because they're so budget and time constrained that's all the resources they had to throw at it. So the lab is full of bespoke machines running a variety of very obscure software, much of it on tailored Windows boxes.
So no, "me and my colleagues" didn't "destroy a multi-million dollar project by trying to save a few dollars."
Enterprise edition is exactly what it says on the tin - you need to purchase certain number of licenses, and these licenses are on top of the OEM licenses.
They are useless for a small team doing a research project.
Why are you trying to excuse Microsoft's behavior here? Who could have predicted that Microsoft would have actually done what they did and forced the Windows 10 update on everyone?
No, it's definitely Microsoft's fault. Even Home users should be able to have the expectation of controlling when or if major OS updates are applied. It's not like "doesn't automatically update to Windows 10" is an advertised feature of Professional.
4 years NSA, 14 years Army Signal Corps (routers/switches/radios) with multiple overseas tours, now working in an enterprise financial data center with CCNA/CISSP and a TS/SCI with full scope poly would put me a little ways out of the "completely incompetent IT staff" category.
It was sent to all versions of Windows, as long as they weren't part of an Active Directory domain, and that detection was pretty flakey, too.
If you had a mixed OS environment with a Samba-based domain (or no domain at all, because these machines just take data from lab equipment and shovel it onto network shares), there was a good chance of the update firing off anyway.
FWIW there was a handful of reports of exactly that happening on /r/sysadmin. There was also a ton of domain joined machines that were spammed with the Windows 10 nagware but didn't actually upgrade on their own.
>I am pretty sure that running home versions of Windows in contexts like this is against the Windows TOS.
For Windows, Microsoft restricts features but not what purpose you can use it for if you're talking about Home versions. For Office however, the home version actually has restrictions in the TOS that limit use to only noncommercial purposes. For a standalone Windows box that is only used for some lab equipment and will never be domain joined there's a good chance that there's no technical reason to go with Windows Pro over Home.
In the atmospheric/climatology lab there are a number of computers running very specific versions of Windows (down to specific patches) tailored to really obscure science research software packages, some industry standard, most custom-rolled and tailored to that project. Which means they're very fragile and need to be left alone.
Of course Microsoft blasts Win10 installer files down our very expensive, very slow satellite link, and proceeds to force Win7/8 systems to Win10 which broke a few multi-million dollar experiments in a place that takes MONTHS to get to in the winter. So effectively killed the research.
And yeah, you can argue "well you should've disabled Windows updates etc etc" but in REALITY, not perfect-theory world, you should be able to trust your OS manufacturer not to force-install a major OS revision without your consent. And in the old days of Microsoft, you could. Now they just say "f--- you, we're installing it regardless what you want."
That company can die in a fire for all I care.