This is a strange issue that is occuring on our switching environment. We have a vmware infrastructure that is connected to two 2960 server switches that are stacked.All the servers virtual machines have static entries and are assigned to the correct vlan on a vswitch. when a reboot occurs the server vm sometimes loosing connection and when we do an ipconfig we see a 169 ip on the host address, the mask and gw are the same as the static entry. Note that after checking the static ip is still there. What we have to do is to set it as another ip and swap to original to get the ip address to take effect.
Besides the vmware infrastructure, I have noticed this issue with clients directly connecting to 2960 poe switches.
Any idea why this may be occuring. initially i was thinking that the issue is related to the vswitch but its occuring to all clients connected to the switches.
I found the issue in our case was related to ip device tracking. I put in the probe delay of 10 like recommended and issue went away.
Are the server's static IPs manually configured or are you reserving them in DHCP ?
Same question for the clients.
All clients which include the server vm and win 7 machine are all set with static ip addresses. This is very strange to me and it is making me think that it may be switch relating.
Can you post the configuration of one of the ports on the switch connected to a device where you are seeing the issue ?
Edit - sorry but just to confirm then, you are not running DHCP at all for any of these devices ie. no reservations within a DHCP scope ?
The server vm was configured on a network that was set for a dhcp scope however the static ip was set outside of that range. So for this server it was set on a network 10.100.0.0/22 which had a dhcp scope for 10.100.0.50-10.100.3.254. I have set the host on the 10.100.0.21 which is outside the scope of dhcp pool.
switchport mode access
switchport access vlan 200
spanning tree portfast
Do you have VMware tools installed on these VMs? Also what vNIC driver are you using?
I'm assuming on your trunk ports coming from your LAN Network that are corrected to your ESXi host, have all the vlans yo need on the trunk as allowed?
Intel Pro-1000 MT is the vnic used by the vm. However it is happening with local machines such as laptops and with differnt os. I am confused here and starting to believe it might be an issue on the switches.
Vmtools is installed. Trunks ports are configured
Did you end up finding your culprit? We are fighting a similar issue here that I also think is upstream in the switch stack (6500) but still troubleshooting everything. Static IP, no DHCP, DHCP is not even enabled or present in this particular /24. Originally I thought we could just ip device tracking probe delay - however quickly learned this is not enabled. Also I know you can modify the registry of the windows VMs but still will get messy with a mass amount of machines.
Configuration from a switch and ESXi perspective looks sound. The VM will reboot over and over using the same physical uplink each time successfully (I can see what physical nic it is using in ESXTOP). Then during my testing, after the say 10th reboot (for example sake) we lose connection. Once I disconnect/connect the NIC again in VMware up it comes. I then immediately see it is now using the other uplink... We have a pair of nics in an active/active port group, one on each switch in this stack for redundancy.
Very odd - if anyone has advice or the OP found his answer would love to hear about it...