Brocade Fabric Watch Administrators Guide (Supporting Fabric OS v7.3.0) - Manual (Page 83)

Refer to the Brocade Network Advisor User Manual for detailed instructions on how to add, edit, view,
and remove thresholds.
Port health and CRC monitoring
There are two types of CRC errors that can be logged on a switch; taken together they can assist in
determining which link introduced the error into the fabric. The two types are plain CRCs, which have
bad end-of-frame (EOF) markers and CRCs with good EOF (crc g_eof) markers. When a crc g_eof
error is detected on a port, it indicates that the transmitter or path from the sending side may be a
possible source. When a complete frame containing a CRC error is first detected, the error is logged,
and the good EOF (EOFn) is replaced with a bad EOF marker (EOFni). Because Brocade switches
forward all packets to their endpoints, changing the EOF marker allows the packet to continue but not
be counted.
For thresholding and fencing purposes, only frames with CRC errors and good end-of-frame markers
are counted. This enables you to know exactly how many errors were originated in a specific link.
Recommended port configuration settings
The following table lists the recommended settings for physical port, E_Port, FOP_Port, and FCU_Port
for both the host device and the storage device.
Recommended configuration for the port class TABLE 22
E=Error_Log, S=SNMP_Trap,
P=Port_LOG_LOCK,
M=EMAIL_ALERT, F=Port Fence
Trait Configuration
Class Area Default Custom Unit Time
Base Low
Thresh High
Thresh Buffer Default Custom Below Above
Port Link Loss X Errors Minute 0 500 50 X
Sync Loss X Errors Minute 0 500 50 X
Signal Loss X Errors Minute 0 5 0 X
Protocol Error X Errors Minute 0 5 0 X
Invalid Words X Errors Minute 0 25 0 X E
Invalid CRCs X Errors Minute 0 5 0 X E
RX
Performance X Percentage Minute 0 100 0 X
TX
Performance X Percentage Minute 0 100 0 X
Port health and CRC monitoring
Fabric Watch Administrators Guide 83
53-1003142-01