beautypg.com

Keep-alive timers and hold-time, L2 protocol forwarding – Brocade Multi-Service IronWare Switching Configuration Guide (Supporting R05.6.00) User Manual

Page 653

background image

Multi-Service IronWare Switching Configuration Guide

623

53-1003036-02

About Multi-Chassis Trunk (MCT)

18

Brocade(config-cluster-TOR)#keep-alive-vlan 10

Syntax: [no] keep-alive-vlan

The <vlan_id> parameters specify the VLAN range. Possible values are 1 - 4090.

When the CCP is down:

If keep-alive-vlan is configured, then CCRR messages are sent periodically for every 1 second
over that VLAN.

When CCP is down and keep-alive vlan is configured, Master/Slave selection is based on
following criteria:

1. If one node’s CCEPs are up and other node’s CCEPs are down then the node with local

CCEPs down becomes Slave

1. Otherwise, the node with higher RBridge ID becomes Slave.

If no packetsare received from the peer for a period of 3 seconds, then the peer box is
considered down.

If keep-alive-vlan is not configured and both the peers are up, then both peers keep forwarding
the traffic independently.

Keep-alive timers and hold-time

To specify the keep-alive timers and hold-time for the peers, enter a command such as the
following.

Brocade(config-cluster-TOR)# peer 1.1.1.1 timers keep-alive 40 hold-time 120

Syntax: [no] peer timers keep-alive hold-time

The parameter should be in same subnet as that of cluster management interface.

The parameter can be 0 to 21845 (default 30 seconds.)

The parameter can be 3 to 65535 (default 90 seconds) must be at least 3 times the
keep alive time.

NOTE

Keep-alive-vlan and keep-alive timers are not related. The keep-alive timer is used by CCP.

L2 protocol forwarding

MCT will forward or drop L2 protocol packets when corresponding features are disabled. The
packets will either be forwarded as regular multicast that floods to the VLAN or dropped. When
forwarded, the packet received from ICL will not be forwarded to CCEP port if the peer MCT switch
has the reachability to the same cluster client.

When designing a network, the ICL port or LAG must have enough bandwidth to support all the
traffic from clients (in case of client links connected to one node failure case).

For L2 forwarding, appropriate CAM profiles need to be used to be able to program all the MAC
entries into the CAM (especially when using LAG interfaces on MCT nodes for ICL and client
interfaces).