Brocade Network OS Administrator’s Guide v4.1.1 - Manual (Page 129)

• Access ports that face servers or hosts. These can be port-channel or physical interfaces, depending
upon the host or server configuration.
• Uplink interfaces that connect to the core (port-channel 6144 in our example topology).
• Interfaces supporting the VCS Fabric topology on all core and end-devices.
NOTE
ISL interfaces have flow control enabled by default.
Do the following to optimize reconvergence.
1. Confirm whether flow control is enabled, by using the qos flowcontrol tx rx command on the
interfaces listed above, as in the following example.
sw87# show running-config interface Port-channel 6144
interface Port-channel 6144
vlag ignore-split
switchport
switchport mode trunk
switchport trunk allowed vlan all
switchport trunk tag native-vlan
qos flowcontrol tx on rx on
no shutdown
!
sw87#
2. To enable flow control on an interface that does not have it enabled, use the qos flowcontrol tx rx
command.
3. Where servers are connected to cluster leaf nodes, is recommended that Link Aggregation Control
Protocol (LACP) vLAGs be used to minimize traffic loss.
Maintaining the VCS Fabric
During a VCS Fabric maintenance window, it is recommended that you do the following.
1. Download the required firmware on all VDX cluster nodes by using the nocommit, noreboot, and
coldboot options as appropriate. The last option applies to non-ISSU firmware downloads.
CAUTION
Do not use the coldboot option unless directed to do so by Brocade Technical Support. If you
do not select the nocommit option, firmware is downloaded automatically. This prevents you
from backing out of an upgrade should that become necessary. Refer to
Restoring firmware in
the VCS Fabric on page 131.
NOTE
In this test topology, because the upgrade is from 3.0.1c to 4.1.1, 4.1.1 is loaded on all VDX cluster
nodes. Your versions will vary accordingly.
2. Verify that there is no control or data traffic outage during the firmware download process.
3. After the firmware completes downloading on all nodes, in large-scale deployments it is
recommended that you reload the fabric principal and multicast root nodes either separately or along
with the "even" nodes.
Maintaining the VCS Fabric
Network OS Administrator’s Guide 129
53-1003225-04