Brocade Fabric Watch Administrators Guide (Supporting Fabric OS v7.3.0) - Manual (Page 18)

Port persistence
The data collected in port monitoring can vary a great deal over short time periods. Therefore, the port
can become a source of frequent event messages (the data can exceed the threshold range and
return to a value within the threshold range).
Fabric Watch uses port persistence for a port event that requires the transition of the port into a
marginal status. Fabric Watch does not record any event until the event persists for a length of time
equal to the port persistence time. If the port returns to normal boundaries before the port persistence
time elapses, Fabric Watch does not record any event.
To set the port persistence time, refer to Setting the port persistence time on page 79.
Port fencing
A port that is consistently unstable can harm the responsiveness and stability of the entire fabric and
diminish the ability of the management platform to control and monitor the switches within the fabric.
Port fencing is a Fabric Watch enhancement that takes the ports offline if the user-defined thresholds
are exceeded. Supported port types include physical ports, E_Ports, optical F_Ports (FOP_Ports),
copper F_Ports (FCU_Ports), and Virtual E_Ports (VE_Ports).
NOTE
Port fencing is not enabled by default. You must manually enable port fencing. Refer to
Port fencing
configuration on page 80 for instructions.
When a port that has exceeded its user-defined thresholds is fenced by the software, the port is
placed into the disabled state and held offline. After a port is disabled, the user must manually enable
the port for frame traffic to resume on the port.
System resource monitoring
System resource monitoring enables you to monitor your system’s RAM, flash, and CPU. You can use
the sysMonitor command to perform the following tasks:
• Configure thresholds for Fabric Watch event monitoring and reporting for the environment and
resource classes. Environment thresholds enable temperature monitoring, and resource thresholds
enable monitoring of flash memory.
• Use the RAM to configure memory or CPU usage parameters on the switch or display memory or
CPU usage. Configuration options include setting usage thresholds which, if exceeded, trigger a set
of specified Fabric Watch alerts. You can set up the system monitor to poll at certain intervals and
specify the number of retries required before Fabric Watch takes action.
For complete information about system resource monitoring, including setting guidelines and default
settings, refer to
System monitoring using the sysMonitor command on page 89.
Switch policies
Switch policies are a series of rules that define specific health states for the overall switch. Fabric OS
interacts with Fabric Watch using these policies. Each rule defines the number of types of errors that
transitions the overall switch state into a state that is not healthy. For example, you can specify a
switch policy so that if a switch has two port failures, it is considered to be in a marginal state; if it has
four failures, it is in a down state. You can define these rules for a number of classes and field
replaceable units, including ports, power supplies, and flash memory.
Refer to Switch status policy planning on page 93 for information on configuring switch policies.
Port persistence
18 Fabric Watch Administrators Guide
53-1003142-01