Technical Product Specification

Platform Management Functional Overview Intel

Server Board S2600WP TPS

 An IPv6 address is 16 bytes versus 4 bytes for IPv4.

 An IPv6 prefix is 0 to 128 bits whereas IPv4 has a 4 byte subnet mask.

 The IPv6 Enable parameter must be set before any IPv6 packets will be sent or received

on that channel.

 There are two variants of automatic IP Address Source configuration versus just DHCP

for IPv4.

The three possible IPv6 IP Address Sources for configuring the BMC are:

 Static (Manual): The IP, Prefix, and Gateway parameters are manually configured by

the user. The BMC ignores any Router Advertisement messages received over the

network.

 DHCPv6: The IP comes from running a DHCPv6 client on the BMC and receiving the IP

from a DHCPv6 server somewhere on the network. The Prefix and Gateway are

configured by Router Advertisements from the local router. The IP, Prefix, and Gateway

are read-only parameters to the BMC user in this mode.

 Stateless auto-config: The Prefix and Gateway are configured by the router through

Router Advertisements. The BMC derives its IP in two parts: the upper network portion

comes from the router and the lower unique portion comes from the BMC’s channel

MAC address. The 6-byte MAC address is converted into an 8-byte value per the EUI-

64* standard. For example, a MAC value of 00:15:17:FE:2F:62 converts into a EUI-64

value of 215:17ff:fefe:2f62. If the BMC receives a Router Advertisement from a router at

IP 1:2:3:4:1 with a prefix of 64, it would then generate for itself an IP of

1:2:3:4:215:17ff:fefe:2f62. The IP, Prefix, and Gateway are read-only parameters to the

BMC user in this mode.

 IPv6 can be used with the BMC’s Web Console, JViewer (remote KVM and Media), and

Systems Management Architecture for Server Hardware – Command Line Protocol

(SMASH-CLP) interface (ssh). There is no standard yet on how IPMI RMCP or RMCP+

should operate over IPv6 so that is not currently supported.

4.6.5.1

LAN Failover

The BMC FW provides a LAN failover capability, such that the failure of the system HW

associated with one LAN link will result in traffic being rerouted to an alternate link. This

functionality is configurable through IPMI methods as well as through the BMC’s Embedded UI,

allowing for user to specify the physical LAN links constitute the redundant network paths or

physical LAN links constitute different network paths. BMC will support only an “all-or-nothing”

approach – that is, all interfaces bonded together, or none are bonded together.

The LAN Failover feature applies only to BMC LAN traffic. It bonds all available Ethernet

devices but only one is active at a time. When enabled, If the active connection’s leash is lost,

one of the secondary connections is automatically configured so that it has the same IP address.

Traffic immediately resumes on the new active connection.

The LAN Failover enable/disable command may be sent at any time. After it has been enabled,

standard IPMI commands for setting channel configuration that specify a LAN channel other

than the first will return an error code.

Intel order number G44057-007 Revision 1.6