02-04-2013 05:29 AM
I have a problem with HP NIC teaming on a Cluster system with Windows Server 2008 R2 SP1:
The servers are configurated in "Network Fault Tolerance with Preference Order" mode and are on the same switch. Many packets are lost and sometimes node1 fails ping to node2. At each reboot the servers not enter to the domain. Without NIC teaming the problem does not exist.
Do you have any solution?
Thanks a lot
04-29-2013 03:18 AM - edited 04-29-2013 03:24 AM
Some more info would be helpful.
Type of NIC, speed set on NIC's, driver version of NIC and driver version of NCU (teaming driver).
Generally you should also be able to see on the switch if there are problems due to cabling/GBIC's/SFP's
If 10Gbit then the type of SFP/SFP+ is important as well as cable (generally multimode fiber if using fiber)
Some of the teaming modes will not only require the ports being used to be on the same switch, but even on the same shelf of the switch (if large/modular switch). However the NFT mode should be fairly switch-agnostic.
With regards to 2008 R2 cluster, be aware that the cluster will build the various Networks automatically, based on IP-address. E.g. with subnetmask 255.255.255.0 you could set up 2 heartbeat networks 10.10.10.x and 10.10.11.x (not teamed) and a public/client network (teamed) 100.10.20.x/24 (replace 100.10.20.x with whatever subnet your servers is on).
Cluster validation testning with networks only should progress without any errors.