Brocade Fabric Watch Administrators Guide (Supporting Fabric OS v7.3.0) - Manual (Page 75)

• Areas: Link Failure Count, Loss of Synchronization Count
Change the default high boundary from 500 to 15 (per minute) for Link Failure Count and from 500 to
45 (per minute) for Loss of Synchronization Count. Leave the Buffer setting to 0 (the default). Set the
alarm configurations to send alerts to both the error log and SNMP. These settings are the same for
an HBA port or a storage port.
• Area: Loss of Signal Count
Unlike the other areas, take a conservative approach for the Loss of Signal Count area. Change the
default high boundary from 5 to 45 (per minute) and set the alarm configuration to send alerts to both
the error log and SNMP.
• Areas: Invalid Transmission Words, Invalid CRC Count
For these two classes, the high boundary settings are split. For Host devices, keep the defaults of
1000 (per minute) and buffer of 100. For storage devices, tighten the boundaries substantially:
change the default high boundary for Invalid Transmission Words to 80, and change the high
boundary for Invalid CRC Count to 40 (per minute).
Hosts and HBAs reboot so do not set alerts for these devices. Storage devices, however, should not
be rebooting, so you should set the alarm to alert more frequently.
Excessive invalid words or CRCs on F/FL_ports lead to fabric congestion and possible frame drops if
left unchecked; therefore, set the alarm to fence the port. Refer to
Port type: E_Port, FOP_Port, or
FCU_Port on page 79 for instructions. In addition, set the alarm configurations to send alerts to
both the error log and SNMP.
• Areas: Receive (Rx) Performance, Transmit (Tx) Performance
Rx and Tx Performance areas are used to monitor the bandwidth utilization of the device ports in the
fabric. Set the high boundary to 85 percent and the alarms to Above and In-Between conditions. The
same levels should be set on both Host and storage device ports.
NOTE
With the increased use of virtual environments, alerts from device ports are increasing more than
ever in the past. This provides a good gauge as to the overall bandwidth requirement changes and
utilization and could indicate that additional ISL trunks are required.
• Area: Link Reset
The goal of the Link Reset area is to avoid excessive link resets which can cause back pressure in
the fabric. The Link Reset area is new; therefore, recommended settings are not available. Keep the
default settings, monitor the results, and adjust your settings accordingly.
• Area: Class 3 (C3) Discards
Unlike the other areas, take a conservative approach for the C3 Discards area. Use the default
settings and configure the alarms for Above. The goal is to locate issues with the device or its
infrastructure, so monitor the data to help isolate issues. Port fencing is one of the recommended
solutions for isolating issues.
• Area: Trunk Utilization
The Trunk Utilization area is new; therefore, recommended settings are not yet available. Use the
default settings, monitor the results, and adjust your settings accordingly.
• Areas: Primitive Sequence Protocol Error, State Changes
These areas are not used for monitoring; therefore, leave the default alarm settings at 0.
FOP_Port and FCU_Port subclass default settings
Port fencing can only be enabled or disabled for the following areas for the FOP_Port and FCU_Port
subclasses:
FOP_Port and FCU_Port subclass default settings
Fabric Watch Administrators Guide 75
53-1003142-01