DELL N3000 User Manual

Managing a Switch Stack

If a host is in the middle of an exchange with the DHCP server when the 

failover occurs, the exchange is interrupted while the control plane restarts. 

When DHCP snooping is enabled, the hardware traps all DHCP packets to 

the CPU. The control plane drops these packets during the restart. The 

DHCP client and server retransmit their DHCP messages until the control 

plane has resumed operation and messages get through. Thus, DHCP 

snooping does not miss any new bindings during a failover. 
As DHCP snooping applies its checkpointed DHCP bindings, IPSG confirms 

the existence of the bindings with the hardware by reinstalling its source IP 

address filters. 
If Dynamic ARP Inspection is enabled on the access switch, the hardware 

traps ARP packets to the CPU on untrusted ports. During a restart, the 

control plane drops ARP packets. Thus, new traffic sessions may be briefly 

delayed until after the control plane restarts. 
If IPSG is enabled and a DHCP binding is not checkpointed to the backup 

unit before the failover, that host will not be able to send data packets until it 

renews its IP address lease with the DHCP server. 

Figure 9-15 illustrates a stack of three Dell Networking switches connecting 

two servers (iSCSI initiators) to a disk array (iSCSI targets). There are two 

iSCSI connections as follows:
Session A:  10.1.1.10 to 10.1.1.3
Session B:  10.1.1.11 to 10.1.1.1
An iSCSI application running on the stack master (the top unit in the 

diagram) has installed priority filters to ensure that iSCSI traffic that is part 

of these two sessions receives priority treatment when forwarded in hardware.