Troubleshooting guide

32 Advanced System Diagnostics and Troubleshooting Guide

Packet Errors and Packet Error Detection

The slow-path and fast-path categories each have a separate configured threshold and associated action

that occurs at the end of the 20-second window:

• For the slow-path category, the three types of slow-path subcategory reports are tallied and

compared to the configured slow-path subcategory threshold.

• For the fast-path category, the two types of fast-path subcategory reports are tallied and compared to

the configured fast-path subcategory threshold.

If either of these thresholds is exceeded, that window is marked as a “bad” window. For each bad

window, the interpreting and reporting subsystem inserts a Sys-health-check error message into the

system log, indicating the primary reason why the window was marked bad.

The interpreting and reporting subsystem also maintains a running history of the windows it processes.

After it completes the window processing cycle, the interpreting and reporting subsystem examines the

window history list and counts the number of bad windows in the list. The bad window count is

compared to a configured window error parameter value. If the bad window count equals or exceeds

the window error parameter value, the interpreting and reporting subsystem alerts the system health

check system that there is a potential service-affecting condition. The system health check system will

then take the action specified in the

configure sys-health-check command (for more information,

see

“Health Check Functionality” on page 71):

• Insert an error message into the system log

• Send an SNMP trap to the configured trap receivers

• Take an offending module offline

• Take the entire switch offline

• Attempt to perform an automated recovery through the use of the packet memory scanning and

mapping capability

NOTE

The default system health check action is to insert an error message into the system log, but in many

environments, it is advisable to configure the system health check to send a trap to configured trap

receivers to provide faster notification of potentially service affecting conditions.

The interpreting and reporting subsystem is enabled by default, and the default values for the slow-

and fast-path thresholds and bad window counters depend on the switch platform type (see

Table 4).

Table 4: Default Threshold and Counter Values

Minimum Summit “

” Alpine BlackDiamond Maximum

3802 3804 3808 6808/6804 6816

Slow-path threshold 1 2 2 2 2 2 2 254

Fast-path threshold 1 1 1 2 3 6 12 254

Window history 1 3 3 3 3 3 3 32

Window errors 1 3 3 3 3 3 3 32