HP Integrity rx2800 - i2 White Paper page 97

• Understanding Requirements for Failover Clusters:
http://technet2.microsoft.com/windowsserver2008/en/library/
6fe86b30-6820-43a2-8321-3e7b51cbe7631033.mspx
• Network Adapter Teaming and Server Clustering:
http://support.microsoft.com/kb/254101
B.B.1.15 My switch is set up for Switch-assisted Load Balancing (Port Trunking) but my
network adapters are not. Why am I having communication problems?
The switch assumes that all ports are capable of receiving on the same MAC
address/es and will randomly transmit frames for any of the NICs down any of the
links for any of the NICs. If the NICs aren't configured for SLB teaming, they will
drop any frame meant for another NIC. Because of this, the switch should only be
configured for Port Trunking after the SLB team has been created.
If a single server with multiple NICs is connected to a switch configured for Port
Trunking and PXE is being used to deploy the server, communication problems will
most likely prevent PXE from completing a successful install on the server. To avoid
such problems, disconnect all NICs except for the NIC providing PXE support or
remove the Port-Trunking configuration on the switch.
B.B.1.16 What is the maximum number of network ports that can be in a single team?
8 ports
B.B.1.17 What is the maximum number of teams that can be configured on a single HP
server?
16 teams
B.B.1.18 What is the maximum number of VLANs that can be configured on a single
network port or a single team?
64 VLANs
B.B.1.19 Why does my team lose connectivity for the first 30 to 90 seconds after the preferred
Primary port's link is restored?
This may be caused by Spanning Tree. Disable Spanning Tree on the port or enable
the Spanning Tree bypass feature if available (for example, PortFast, bypass).
B.B.1.20 If I make an Altiris image of a server with a team, can I deploy that image onto
other servers?
Yes, but the team's MAC address registry entry will have to be restored individually
on all servers the image was deployed on. A better solution is to use the Windows
Command Line utility called CQNICCMD.EXE (delivered in the NCU driver
download) to import an team XML script to restore a teaming configuration.
B.B.1.21 Why do I still see heartbeat frames after disabling them?
Even when heartbeats are disabled, an HP team must make sure that the team's
MAC address is known by the switches in the network to prevent flooding of traffic
destined for the team. To achieve this, the team needs to transmit a frame periodically.
The frame used for this purpose is a heartbeat frame. Heartbeat frames are also used
during a failover to notify the switch of MAC address changes on the teamed ports.
The purpose for disabling the transmit and receive validation heartbeats on the
Settings tab of the Team Properties page is to either reduce the number of heartbeat
frames on the network or to disable the Path Validation redundancy mechanism.
Disabling transmit and receive validation do not completely disable the transmission
of heartbeats on the network.
B.B.1.22 When should I increase the heartbeat timers for a team?
The heartbeat timers should be increased when heartbeat failures are caused by
latency in the network infrastructure. This is an extremely rare problem.
B.B.1.23 I need NIC redundancy, switch redundancy and Rx/Tx load balancing. What
teaming type should I use?
When NIC redundancy, switch redundancy, and load balancing are all required,
the only option is Dual Channel or Dynamic Dual Channel. An alternative is to use
97