Brocade FICON Administrator's Guide v7.1.0 (53-1002753-01, March 2013)

FICON Administrator’s Guide 69
53-1002753-01
Port fencing
5
Port fencing
Occasionally, bad optics and cables can cause errors to occur at a rapid rate that error processing
and sending and processing RSCNs can cause fabric performance problems. Port fencing allows
the user to limit the number of errors a port can receive by forcing a port offline when certain error
thresholds are met.
The port fencing feature is configured through Fabric Watch. For more information on configuring
Port Fencing, refer to the Fabric Watch Administrator’s Guide.
Defining port fencing
Following is an overview of steps needed to define port fencing.
1. (Optional) Clear all alarms.
2. Define threshold levels.
3. Define alarm action.
4. Activate alarming.
NOTE
Establish a Telnet session with a tool such as Putty that allows the columns to be increased. This is
because some of the displays use more than the standard 80 columns that programs such as
HyperTerminal support. Recommended number of columns is 120.
Settings for FICON environments
For typical FICON environments, port fencing is usually only set for CRC errors and Invalid Words.
The default of 1,000 errors per minute is a little high for CRC errors and Invalid Words. A more
common setting is 50 errors per minute. This is high enough to ignore occasional errors and
transient errors due to re-cabling but low enough to stop problematic optics from causing fabric
issues.
By default, the alarms are set to fence the port, log an alert, send an e-mail, and set an SNMP trap.
In most FICON environments, only fencing the port and logging the alert are desired.
The following are the default thresholds:
• Low 0
• High 1,000 errors per minute
FICON information
You can display link incidents, registered listeners, node identification data, and FRU failures, as
described in the following sections.
Link incidents
The registered link incident record (RLIR) ELS contains the link incident information sent to a
listener N_Port.