Hyper-V servers DHCP

[RahamimL]

Hi everyone,

We upgraded our Hyper-V cluster from 2012 to 2016. (Hardware compatibility issues). Ever since we upgraded, every time a host is restarted, we get a weird DHCP request. We monitor our DHCP server filter, which is installed on a 2012 R2 OS, with SCOM to check if an unknown MAC address is attempting to request an IP. I ran Wireshark and I found the following:

After the server is drained, paused and restarted, a request is coming to the DHCP server with an unknown MAC address. the source of the request is from the host's physical MAC address which has a static IP address. The client IP address is the Host's IP. The client MAC address is a random address which is not a part of anything we setץ

I even tried installing SCVMM 2016 and had the same issues. The only thing that is weird is, when a VM is restarting I don't have this issue.

I have 2 clusters in 2 different locations with the same problem.

Any help would be appreciated.

Rahamim.

[#2 · Accepted solution]

Found the culprit!!!

First of all, When installing Hyper-V and using an ISCSI LUNs, the MSiSCSI services is throwing DHCP discover packets when it start. After it finds the NIC that uses the iSCSI targets it uses that NIC and doesn't attempt to search for DHCP until the next reboot \ restart of the service.

Apparently when installing SCVMM server and SCVMM agent this is what happens: On the SCVMM server the iSCSI initiator service is installed and starts running (like the Hyper-V) which causes the first DHCP discover packets. Than, On all the server that SCVMMAgent service is installed, a DHCP discover packet is broadcasted repeatedly in a constant interval for each server.

The solution: https://download.microsoft.com/download/A/E/9/AE91DEA1-66D9-417C-ADE4-92D824B871AF/uGuide.doc
Although this is for server 2008 and 2003, when Creating the MSiSNS Key and adding the DHCPRetries DWORD with the value 0, it causes iSCSI to stop sending DHCP discover packets.

Hope this helps anyone.

Rahamim.