PCA Forums

So i'm having an issue at work. Here is my network setup...

Server 1 - Primary Domain Controller, win2008r2 x64, using hyperv, all power saving options turned off including unchecking 'allow this device to be turned off to save power' in device manager, static IP, DNS, and Gateway, running Firebird SQL database, also used as a file server
Roles- DNS Server, DHCP server, Active Directory, Hyoer V


Server 2 - Win2008r2, using hyper-v, all power options turned off including nic setting as obove, static IP, DNS, and gateway
Roles - DHCP server, hyper -V, Remote Desktop Gateway, IIS, 3cx phone system, MS SQL server

So here is what happens. At night, we lose all network connectivity. Network connectivity only returns when I initiate a remote desktop connection to one of the servers. Its the strangest issue. The weird part is that it also takes down the mapped drives on server 1, as well as everything on server 2. This leads me to believe both servers are going down. I am using a Cisco Small business switch. Is it possible the switch is going to sleep??

Seems more likely it's the switch. Have you power cycled it? Can you swap it out?

Process of elimination. If that doesn't fix it, then it's likely an issue with both servers which seems unlikely, unless maybe some NIC driver update they both got is causing it.

Speaking of which, look back through the Windows Update logs and see if this coincided with an update.

anything special going on at night (such as backups that hit that switch) that would be different from the daytime traffic? if it's only happening at night, something must be triggering it. server power settings don't go to sleep by default, so it shouldn't be that. there was a 2008 nic sleeping issue that i read about a while back, but not sure if R2 was affected. i'd also think it's the switch and something like backups are hitting it too hard, then it locks up or similar.

one other thing i thought of would be the arp cache on the servers. when it happens (if possible to do this from the server), do an "arp -a" at a command prompt and verify that the MAC addresses and IPs all look ok. i've heard of duplicate MAC addresses occurring (don't ask me how) and when a server's NIC does a lookup and populates the arp cache, it gets more than one response with the same MAC, so it fails. this could possibly happen due to your use of Hyper-V as well, since the dynamic MAC pool being handed out to the VMs could overlap. unlikely, but worth a shot. check the VM's MAC addresses on each server.

.02

Thanks for the replies. I have a new switch on order to test it. This is what I'm leaning towards.

It's not every night. And there are no network backups, only local to an external hard drive on each server followed by crash plan pro subscription.

Next time it happens I'll try the arp-a command. It's also unlikely to be a driver issue as they both use different brand nics

any luck with the testing?

I just replaced the switch on Monday. So far so good. I'll keep you guys posted through the weekend.

Everything is still running fine. The real test is the weekend!