Installation Guide

Table Of Contents
Enables or disables auto-tuning on TM for burst CRC.
NOTE
Default: Enabled.
The following example demonstrates the CLI command necessary to execute the sysmon fe auto-tune command for enabling high
rate auto-tuning on FE for burst CRC.
devicedevice# sysmon fe auto-tune
The following example demonstrates the CLI command necessary to execute the no sysmon fe auto-tune command for disabling
high rate auto-tuning on FE for burst CRC.
device# no sysmon fe auto-tune
The following example demonstrates the CLI command necessary to execute the sysmon tm auto-tune command for enabling high
rate auto-tuning on TM for burst CRC.
device# sysmon tm auto-tune
The following example demonstrates the CLI command necessary to execute the no sysmon tm auto-tune command for disabling
high rate auto-tuning on TM for burst CRC.
device# no sysmon tm auto-tune
Fabric link monitoring
Fabric monitoring will normally shutdown the link which was put down by hardware during burst CRCs. With auto tuning enabled, the link
gets tuned for the first time when the link goes down, and if the same link goes down even after tuning, it gets powered down by fabric
link monitoring.
Fabric link monitoring is running on MP only, and relies on the built-in hardware feature of the fabric element (FE) chipset sitting on the
switch fabric. This hardware feature tracks the leaky bucket value. Initially, the leaky bucket value is set to the value of 63. For each cell
with CRC, the FE will decrement the leaky bucket by the value of 1.
When the leaky bucket value is below the DOWN threshold, the FE marks the link as DOWN. For every 256 good cells received, the FE
increment the leaky bucket value by 1 until it reaches the maximum value of 63. When the leaky bucket value passes the UP threshold,
the FE marks the link as UP. The DOWN and UP thresholds are as follows:
DOWN threshold = 16
Leaky bucket < 16 change from UP to DOWN status.
UP threshold = 32
Leaky bucket > 32 change from DOWN to UP status.
NOTE
In certain scenarios, the link can be alternating UP to DOWN and DOWN to UP. When this happens, there is traffic loss in the
system. To avoid this scenario, software fabric link monitoring will monitor the link status every second. The link is tuned once
after the link is DOWN >= 10 times after a 20 second monitoring period. After tuning the link, if the link is DOWN again due to
CRCs, then the link is powered down. However, the link is not powered DOWN with auto-tuning enabled; instead, the link gets
tuned for the first occurrence of DOWN.
Forward Error Correction mode
Using Forward Error Correction (FEC) mode enabled modules on a Brocade MLXe series chassis will reduce packet drops due to CRC
errors. FEC will automatically be enabled on supported line cards and fabric links in a Brocade MLXe series chassis.
Product Overview
Brocade NetIron MLXe Series Hardware Installation Guide
53-1004203-03 75