We have HP Proliant DL160 G5 servers connected to HP 2524 100Mbps switches at remote site locations and those switches are running the latest software F.05.70. The servers are running Server08 Ent SP1 with Hyper-V and each are hosting 3 virtual machines--all Server 2003 SP2.

The servers have onboard NC105i Gb adapters and  NC360T dual port Gb adapters. One of the onboard NC105i adapters is used for the Hyper-V host management and the remaining NC105i and both of the NC360T adapters are teamed for use of the VM virtual network traffic. All adapters have the very latest drivers and Network Configuration Utility software.

We have been seeing randomly occurring periods of packet loss when we persistently ping the host management addresses of the remote site Hyper-V host servers. The packet loss seems to last for 1-2 minutes and occurs maybe 2-3 times in a 24 hour period—mostly but not exclusively in periods of lower traffic. We have turned off power mgmt on the NICs both on the adapters themselves and within the OS which is Server08 ENT SP1 64-bit. Less frequently, we see periods when the connection to the virtual machines also is down for a similarly brief period of time. The reason we see this packet loss is that we use network monitoring software called ActivExperts which pings the host mgmt interface every minute and reports an email alert when that ping is unsuccessful on two successive polls.

Both the server adapter properties and 2524 switch port settings are set to auto speed and duplex. The Teamed adapter settings for Team Type and Load Balancing method are both set to Automatic (recommended). The switches are currently showing the teamed adapters on their 3 ports as dynamic lacp aggregates.

Based on some advice from MS Technet Hyper-V blogs we have turned off offloading in any form (TCP/IP, large send, etc.), disabled the Server08 network “auto tuning” via the netsh command and also actually disabled the virtual switch NIC having a dhcp address which Hyper-V creates but none of these settings have addressed the packet loss issue. I also disabled power mgmt on each NIC and in the OS. I am also providing a link to the Hyper-V thread I created:

http://social.technet.microsoft.com/Forums/en-US/winserverhyperv/thread/99020805-5e76-461f-b16d-f3f56c5438ef

I had received guidance re changing the HP2524 settings to create manual trunks, turn off lacp and manually set all ports to 100-Full:

These are my questions based on that guidance: 

1) Do I go through the procedure of creating the trunks, turning off lacp and setting manual 100/full on each of the switch ports to which the Hyper-V host server teamed adapters and host management adapter are connected first and then afterwards modify the settings on the adapters on the server? Or do I start on the server adapter setting first and then do the switch settings?
 
2) Will I have to dissolve the adapter team on the host servers and make each individual teamed adapter 100/full?

3) If so and after I do that, when I recreate the team, these are the choices I have on the properties of the NIC team for Team Type Selection and Transmit Load Balancing Method.  I rather suspect that it makes a big difference which selection is made. Which selection should I make for Team Type Selection and Transmit Load Balancing Method:

Team Type Selection Choices:

Automatic (recommended)
802.3ad Dynamic with Fault Tolerance
Switch-assisted Load Balancing with Fault    Tolerance (SLB)
Transmit Load Balancing with Fault Tolerance (TLB)
Transmit Load Balancing with Fault Tolerance and Preference Order
Network Fault Tolerance Only (NFT)
Network Fault Tolerance with Preference Order
 
Transmit Load Balancing Method:

Automatic (Recommended)
TCP Connection
Destination IP Address
Destination MAC Address
Round Robin (Packet order not guaranteed)

4) Am I even asking the right questions to fix this packet loss issue, or is there some other factor I have missed entirely?