2008R2 SP1 Hyper-V Server losing IP address - DHCP renew failing
Hi, I have a Windows 2008 R2 server running as a VM on Hyper-V. It's been running fine for some time until a couple of weeks ago when it seems to have started losing it's IP address every 4 days. This may coincide with installing SP1 on it. Basically one day it will be fine, the next it's fallen off the network, unpingable etc. Logging onto the machine I discover that it's DHCP assigned IP is gone, replaced by a 169 autoconfiguration address. Disabling and re-enabling the network adapter it still gets a 169. Ipconfig /renew gives the error "An error occurred while renewing interface...An attempt was made to access a socket in a way forbidden by its access permissions". Rebooting the server gets it back to a DHCP assigned address which works fine for 4 days. My DHCP lease is set to 8 days, so I assume that it's trying to renew after 50% of this time and failing. The VM is running on a Hyper-V failover cluster. Other machines on the cluster are working fine. DHCP supplied by a 2008R2 SP1 server, not seeing DHCP issues elsewhere on the network. Things I've tried already - netsh int ip reset netsh winsock reset Removing and re-adding the virtual network adapter Checking event logs - can't see anything related SFC / Scannow So if anyone has any ideas what else to try, please give me a shout. By the way, I'm not going for the obvous solution of a static IP as the Hyper-V cluster is spread over two subnets so for failover purposes I'd prefer the IP to be dynamic. Thanks, Tim
March 1st, 2011 6:05am
Ok, I've done a little more reading about this and came accross this http://support.microsoft.com/default.aspx?scid=kb;EN-US;974909 Apparently there is a known issue where Hyper-V clients can lose connection under high network load. The description is a little vague so I'm not 100% sure this is the same issue I'm seeing, but I might try the hotfix. This also got me thinking as this server is running a network monitoring tool - PRTG Network Monitor http://www.paessler.com/prtg so is going to be making a large number of outbound network connections. I've tried disabling PRTG and the problem with the ipconfig /renew goes away. I'll raise this with the software vendors to see if they have any thoughts. The puzzle remains what caused this problem to start happening, was it SP1. The hotfix predates SP1 so is it recommended to apply it to SP1, have the changes in the hotfix been incorporated in SP1? Edit looking at the file versions in the hotfix, they are considerably older than the ones present on the VMHost machine which is on SP1, so I'm reluctant now to try the hotfix. Tim
March 1st, 2011 8:27am
I would not recommend running Windows 2008 R2 server with non-static IP. However, if you choose to, appling a static will determine if DHCP address is expiring and not being renewed.W.Maxx MCITP MCSE MCSA MCP 2k8 2k3 2k nt4 AD, Exchange, DNS, msProject, SharePoint IBM iSeries RIM blackberry
March 1st, 2011 11:03am
Hi, Thanks for the post. Please understand that the Hotfix KB974909 has been included in Windows Server 2008 R2 Service Pack 1. You said that the ipconfig /renew command could work properly if we disabled PRTG. I am wondering if the losing IP issue will occur if we disable PRTG. Please try to shorten the DHCP Lease Duration and verify it. Meanwhile, please type ncpa.cpl in Run box. In the Network Adapter properties, uncheck the items regarding the third party programs for test. Thanks, MilesPlease remember to click Mark as Answer on the post that helps you, and to click Unmark as Answer if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
March 2nd, 2011 4:12am
Hi Miles, I followed your suggestion and reduced the DHCP lease time, the problem with losing the IP address does not occur when PRTG is disabled. As the start of this problem seemed to coincide with installing SP1 I've uninstalled this from both the VM and the Hyper-V host machine - but the problem is still there. I've checked the network adapter properties and there are no third party components in there. All in all it's looking more like an issue with PRTG now, I'll see what their support have to say about it. Tim
March 2nd, 2011 8:25am
Hi Tim, Yes, there must be something wrong with PRTG. Please let us know if there is any update on this issue. Thanks, MilesPlease remember to click Mark as Answer on the post that helps you, and to click Unmark as Answer if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
March 2nd, 2011 9:51pm
Their response was that they had not come accross the issue before, but it was very unusual to have somone running a server on a dynamic IP, somthing which they advise against. Anyway, I've now put the server on a static IP which hopefully will resolve the problem. Thanks, Tim
March 4th, 2011 11:50am
I don't think DHCP is the cause of this issue. I have a 2008 R2 non-SP1 guest VM with a static IP running Sharepoint 2010 on a 2008 R2 SP1 hyper-v cluster that is having this same issue. The guest randomly loses network connectivity after a few days. Initially, I was rebooting the guest, but found I can live migrate the guest to another node and it gets network connectivity back. This VM has been running for months in this configuration without issues - until applying SP1 to the hyper-v host nodes. Update: After further troubleshooting I realized that I hadn't upgraded the integration components on the guest after applying SP1 on the host (somewhat embarrased I missed this). This was the only guest having this problem and I realized I upgraded the other guests on the cluster to SP1, so they automatically got the new version of the IC. I think this very well may be the issue, but time will tell.
March 9th, 2011 12:14pm
I don't think DHCP is the cause of this issue. I have a 2008 R2 non-SP1 guest VM with a static IP running Sharepoint 2010 on a 2008 R2 SP1 hyper-v cluster that is having this same issue. The guest randomly loses network connectivity after a few days. Initially, I was rebooting the guest, but found I can live migrate the guest to another node and it gets network connectivity back. This VM has been running for months in this configuration without issues - until applying SP1 to the hyper-v host nodes. I think the issue is with SP1.
March 9th, 2011 12:14pm
I have had something simular with a 2008R2 VM running on Hyper-V. I am not sure what has caused it, but the machine has a Static IP and runs AD etc. We had some strange network behaviour that looks almost like an internal DOS attack and after re-starting the VM, it had a DHCP assigned IP address. I had to fight for a long time to get hte static IP up again.
March 16th, 2011 7:25am
Same issue here on 3 Hyper-V 2008 R2 SP1 servers. Contacted MS Support and this issue was confirmed. They are working on a patch for this.
March 31st, 2011 3:55am
We have 6 2008 R2 SP1 VM guests with a static IP on a 2008 R2 SP1 hyper-v cluster (2 nodes). The guests randomly loses network connectivity. The network card is in disabled state and it cannot be enabled again unless the guest is rebooted. These VM's have been running for months in this configuration without issues before applying SP1 to the hyper-v host nodes and guests. To be sure I will reinstall the integration services again on all VM's and I will let you know what the results are. Also I will open a call with MS Support.
March 31st, 2011 7:50am
I can confirm that installing the integration components DID NOT resolve this issue. I've had the issue occur again since doing so. Robin, please update with a hotfix # if/when you become aware of one.
March 31st, 2011 12:27pm
I can also confirm reinstalling the integation components does not fix the issue. To make everybody happy: Microsoft has released a hotfix that solves this problem http://support.microsoft.com/kb/2263829/en-us
April 7th, 2011 3:24am
I have the same symptoms, however it usually happens with the Windows 7 VM's. Both have SP1, however the issue seemed to happen occasionally before SP1. The VM loses the DHCP address and has an auto config address. However it still has the subnet, default gateway, dns, etc from my dhcp server. Logging on to the console and running "ipconfig /renew" on the address gives an error, but 10 secs later it has the ip address back. My Win2008 R2 Hyper-V server doesn't have SP1 installed yet either. This issue is really sporadic. Sometimes it'll happen a couple times a day, then i'll go for weeks without an issue. Do i have to setup my VMs so they have static ips so I don't have this issue?
June 7th, 2011 4:00pm