Windows 10 loses some internet and browser connectivity about 24 hours after reboot

@alex,

It does appear to be solved. I believe it was actually an issue where TCP ports were being exhausted and not related to DNS at all. I did some workarounds as referenced below and it was much better. Then there was a Windows Update about a month ago that seemed to do away with the issue entirely. So I think it may have started from an earlier Update somehow as I suspected.

Similar to this:

https://msdn.microsoft.com/en-us/library/aa560610%28v=bts.20%29.aspx?f=255&MSPPError=-2147217396

But I'm running Windows 10 and I don't think this directly applies. I have tried to watch netstat before and after it breaks and it doesn't appears to have an unruly amount of ports used. The symptoms are consistent with port issues though:

• Existing connections typically remain open for a period of time (music streaming, existing SQL Server query connections, skype, slack, etc)
• New connections are blocked (browsing, new SQL connections, etc)
• Shutting down services that are using a lot of TCP connections (like Coldfusion server) will temporarily restore service for some types of connections (some browser requests, etc)

I have watched my vshub.exe error logs from Visual Studio and see that it's cycling through a bunch of ports trying to open a connection and it's constantly blocked (49152, 49153, 49154, ...). However, those ports do not show up under netstat as blocked or in use at all. They're simply not listed. But vshub still cannot bind to the port.

But this is what I actually did to make it much better:

https://deploymentresearch.com/Research/Post/532/Fix-for-Windows-10-exhausted-pool-of-TCP-IP-ports

I settled on these options (listed in the article):

TcpTimedWaitDelayREG_DWORD: 0000001e (30) (hex) 

MaxUserPortREG_DWORD: 0000fffe (65534) (hex)

TcpMaxDataRetransmissionsREG_DWORD: 00000005 (5) (hex)

I didn't end up setting TcpNumConnections at all.

Hopefully that helps! If it doesn't and you haven't done the latest comprehensive Update, I would definitely try that as well.

That makes some sense on the networking side. Like it's possibly hitting some "buffer" or something that fills up and only gets reset on reboot.

I know that the static IP addresses remain connected to the machine. And I'm 99.9% sure that I ran a test to ping my machine from other machines on the network after the connection is lost. I can also access other machines on my local network (including mapped network drives on my NAS) when it happens. But the browser based connections are not accessible.

So it's like 80/443 stop working correctly but most other connections remain. Which screams either virus, malware, or firewall to me but I can't seem to find any apps or processes that are doing it.

I do see some events in the Event Viewer of the following (every 30 minutes or so):

The processing of Group Policy failed because of lack of network connectivity to a domain controller. This may be a transient condition. A success message would be generated once the machine gets connected to the domain controller and Group Policy has successfully processed. If you do not see a success message for several hours, then contact your administrator.

I also see this every 90 minutes or so this morning:

The system failed to register host (A or AAAA) resource records (RRs) for network adapter with settings:

           Adapter Name : {22501B19-[MASKED]- 404B0A9BA624}

           Host Name : [MASKED]

           Primary Domain Suffix : domain.[MASKED]

           DNS server list :

                 8.8.4.4, 8.8.8.8

           Sent update to server : <?>

           IP Address(es) :

             [My 4 Static IP's listed here; MASKED]

The reason the system could not register these RRs was because the DNS server contacted refused the update request. The reasons for this might be (a) you are not allowed to update the specified DNS domain name, or (b) because the DNS server authoritative for this name does not support the DNS dynamic update protocol.

To register the DNS host (A or AAAA) resource records using the specific DNS domain name and IP addresses for this adapter, contact your DNS server or network systems administrator.

And one last weird one:

Initialization failed because the driver device could not be created. Use the string "000000000100320000000000D71000[MASKED]1000000000000000000000000000000" to identify the interface for which initialization failed. It represents the MAC address of the failed interface or the  Globally Unique Interface Identifier (GUID) if NetBT was unable to  map from GUID to MAC address. If neither the MAC address nor the GUID were  available, the string represents a cluster device name.

I could try doing a BIOS update from Dell (Precision T5610) but again, it's just weird that it was working fine for 3 years and then some recent change causes it to freak out every day. And it's not the same time every day. It's literally about 24 hours each time. I did have it once over the weekend take somewhere between 24-48 hours but otherwise it's very consistent.

Badly made or misconfigured drivers can cause issues like this, such as after so many packets or hours running they can hit the issue and be unable to recover. I had a similar frustration with my Dell Venue 8 Pro but it hasn't happened since I updated the BIOS, although I can't say for absolute certainty if that was the fix.

I'd still encourage you to get the drivers from the system builder/configurator like Dell/HP/Acer/etc, since they will often modify the driver from Intel and ship their own configured version.  But considering an entire different network card connected by USB has the same issue, I'd be inclined to agree that it isn't the fault.

When you lose connectivity, does your IP address become invalid?  Or perhaps, does the Device Manager entry for the network card still report that the device is working normally?

@Shawn, Thanks for the reply. I should have been more clear in my "reset winsock" line item to say that I did run those commands as Admin in command line. I went ahead and updated the initial post to reflect that. The end step of those commands is always to reboot. Which of course solves the problem temporarily (as does any reboot) but then it reoccurs approximately 24 +/- hours from that point.

As far as the drivers go, I had an Intel card onboard that I tried both the Device Manager as well as going directly to the Intel site to get the latest drivers. Neither worked. So I tried the "generic" card I had installed years ago and got the same result. Then I went and bought a brand new USB network card from BestBuy in hopes that a different method altogether might provide a positive result or at least some different troubleshooting data. Unfortunately it worked exactly the same as the other two cards I had tried.

And I'm wondering why the connection would reset with every reboot and then break after an entire day if the drivers were in fact bad (not questioning you but the machine issues themselves). It's obviously a possible solution that needs explored, but it will work without fail after a reboot. So I'm not convinced that drivers would necessarily fix it even though I have tried several in order to correct this with 3 different cards.

If you have any other thoughts please let me know.

If you only updated drivers from Device Manager, that update check isn't very good. Companies have to voluntarily submit drivers to get found by that tool, and even if there are they are usually outdated.

I'd at minimum re-apply whatever driver is on the OEM's website for that model. Even if it's the version you already have I'd still do it.

I know you've tried a ton, forgive me if you did this, but I found success with it before:

1. First, open an admin-mode command line interface:

On the Start screen, type cmd and press Ctrl+Shift+Enter. You'll be prompted for this admin action.

Make sure the top of the command prompt window says "Administrator" in the title. 2. Type these two commands to reset the networking system:

Type each of these lines, press enter after each one

netsh winsock reset catalog

netsh int ip reset reset.log hit

After running both of these, hopefully everything can connect again. Let me know how it goes.