Hyper-V sets VM created date to 1601, in the reign of Good Queen Bess Virtualisation mavens have been reminded that Microsoft’s Hyper-V has a bug that occasionally resets the “created” date of virtual machines to the year 1601. That's 1601 as in during the reign of Queen Elizabeth the first. Which we feel compelled to point out was quite a few years before the server was invented. The bug's …
I want to know if the kids with Windows Touch 10 can setup the Server via a KVM with Server 2016. I really gotta know as this is mission critical that the desktop/laptop/tablet/fablet/phone/XBox interface make it so we can use those same apps on our servers too.
I'm guessing you're trolling, but with 2012R2 it's quite possible to administer it from the console via KVM, or via Remote Desktop, or via Powershell.
Mind you, the HyperV console is clearly designed for keyboard and mouse, so you might have trouble using a touchscreen for that.
Youngsters may not realize that computer dates are counted from 1 January 1601. From the almighty Wiki: "1601-01-01 is used as the base of file dates and of Active Directory Logon dates by Microsoft Windows. It is also the date from which ANSI dates are counted and were adopted by the American National Standards Institute for use with COBOL and other computer languages. ... All versions of the Microsoft Windows operating system from Windows 95 onward count units of one hundred nanoseconds from this epoch."
And to atone for posting that, from Twain's 1601: "In ye heat of ye talk it befel yt one did breake wind, yielding an exceding mightie and distresfull stink, whereat all did laugh full sore."
No, that was a very stupid idea, bright only in the narrow minds who believe there is a "before Unix" and "Anno Unix" era (included its doomsday in 2038...)- .. not that 1601 is far better, but at least lets compute date before 1970 a little easier. But I bumped into that choice too when writing an application for a museum... it has items older than 1601.
A better choice would have been using Julian Date, which goes far back enough to allow for most needs.
"A better choice would have been using Julian Date, which goes far back enough to allow for most needs."
Actually the FILETIME type is signed and has several dozen millenia on either side of the 1601. I assume 1601 was chosen because it falls on a 400-year boundary. However, 2001 would have worked fine and had the additional advantage of placing the epoch firmly within the era of atomic clocks. I assume the historic date was chosen because programmers have inherited an irrational fear of negative dates and times from society as a whole. They should have consulted an astronomer (in which case they'd have ended up measuring from 2000.0).
Just right off the top, and something I'll look at, is the data all in the NVMe and virtual machines each doing their own thing without having to look at a reduced map of the data prefetch. Probably something wrong with that, totally off the cuff.
What I'll be watching is what sort of tax nVidia attaches to be the "minimum GPU" card for SR-IOV here. I've the NVMe, Intel network cards (chosen for the role), and probably my SAS controller to boot (literally). That baby I usually have attached to a bank of SSDs.... Anyway having got the download last night, time to play.
"It's Christmas!"
So I used it in Windows VMs running on top of Linux kvm hypervisor since 2014, thanks to work of Alex Williamson and others on vfio, for both GPU cards and USB controllers. It has also been available and supported on Xen and VMware. Now Microsoft is catching up, I wonder if they also implemented proper DMA isolation with IOMMU (or similar). Also I wonder if they support GPU assignment which is a little more difficult, because of computer's BIOS (or UEFI) insistence on using screen before hypervisor is loaded.
Anyway, good job Microsoft on catching up. At least a little bit.
But in Protestant countries there might be code needed to handle the "hole" introduced by the conversion to the Gregorian calendar.
http://www.public.iastate.edu/~pfoley/Decleapyear.htm
Now I'm curious how MSFT does account for that. Not curious enough to do any real research, but considering that IIRC, they didn't fix the last "Is 2000 a leap year?" bug in Windows until November 1999...
PCi passthrough is something that I have really missed since migrating from ESXi, I have a number of ESXi hosts as user machines with full access to discrete GPU and local USB ports... live migration isn't possible, but in my case its not needed...
Being able to move those to Hyper-V, especially with the basic Server licence model is a huge boon.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
