Last week, Microsoft pushed an update to Windows 10 that broke DHCP and knocked some users offline until they rebooted their systems. The update is believed to have been part of cumulative update KB 3201845, which was released on December 9. After it was released, multiple European users reported being kicked offline. It’s not clear if the problem was isolated to Europe or not, but Microsoft is displaying a global banner that declares all users with Internet connectivity problems should restart (not shut down) their hardware.
Yesterday, Microsoft released KB3206632, which Ars Technica believes might have fixed the issue. The new patch contains the following note: “Addressed a service crash in CDPSVC [Connected Devices Platform Service] that in some situations could lead to the machine not being able to acquire an IP address.” If you look up the CDPSVC, it’s described as follows: “This service is used for Connected Devices and Universal Glass scenarios.” Connected devices is self-explanatory, but we haven’t been able to find a definition for what “Universal Glass” is.
Either way, the update broke Windows 10’s ability to configure DHCP (Dynamic Host Communication Protocol). DHCP is the protocol that distributes network configuration data to all the relevant devices on the network and handles automatically assigning IP addresses, for example. You don’t need a DHCP server to access the Internet, but most home networks are configured to expect one, and the average user probably isn’t comfortable with the process of mapping out static IPs to each device on the network.
In this case, the problem can be solved with a simple “ipconfig /release” command, followed by “ipconfig /renew”. Some users are also reporting that this is fix is insufficient, and a separate set of commands are also needed, specifically: “netsh int ip reset” followed by “ipconfig /flushdns”. Combined, these should resolve any issues you experience, and allow an affected system to reconnect to the Internet and download the appropriate patch.
The larger issue here, of course, is that these kinds of mistakes have become a regular part of the Windows 10 update process. In the past 12 months, we’ve seen multiple updates that variously bricked systems, broke Internet connectivity, or caused random crashes when ordinary USB devices (Kindles, in this case) were plugged into the system. That’s not even counting the malware-like activity of the last few months of the “Get Windows 10” campaign and the ill-will that caused towards Microsoft.
Every operating system has these kinds of problems from time to time, including previous versions of Windows. This isn’t the first time Microsoft has had to push a patch to resolve issues it caused for itself with a previous update, and this kind of problem occasionally hits Linux and Apple users as well. But even after allowing for all of those factors, Windows 10 seems to have had more problems with weird corner cases, random bugs, and issues cropping up that the company’s Fast Ring / Slow Ring early adopter update system simply hasn’t been able to resolve.
One potential reason for this is the type of OS testing Microsoft encourages its early adopters to engage in. If you’re in the fast ring, Microsoft recommends you not test your primary system and that you test within a virtual machine when possible. There’s a lot of things that can be checked that way, but certain issues — like USB device verification, for instance — probably don’t happen when users are running within a VM.
To date, Microsoft has yet to announce any substantive changes to its policies that would close these gaps.