CRC error monitoring

Note:

This feature is supported on all 7210 SAS platforms as described in this document, except the 7210 SAS-K 2F1C2T.

This feature allows the user to track CRC (cyclic redundancy check) errors received on a specific port and notify them. The detection mechanism is based around a configurable threshold specified by the administrator. Two thresholds are configurable, one for CRC degrade and one for CRC signal fail. The first threshold crossing generates an alarm, log entry, trap, but does not bring the physical port down, while the second (signal fail) threshold crossing logs an alarm, trap generation, and brings the port operationally down.

The thresholds are configurable with the following CLI command config>port>ethernet crc-monitor.

This behavior is enabled on a per-port basis. By default, the command and functionality is disabled for the signal degrade and the signal fail.

The user can configure different values for the sf-threshold and the sd-threshold. However, sf-threshold value must be less than or equal to the sd-threshold value.

The values provided by the user for threshold and multiplier is used to compute the error ratio as (Multiplier * (10 ^ - (threshold value)). Port Stats are collected once per second and accumulated over the configured window size. Each second, the oldest sample is discarded and the new sample is added to a running total. If the error ratio exceeds the configured threshold (the preceding computation) over the window size for two consecutive seconds, appropriate actions are taken as follows:

  • If the number of CRC errors exceeds the signal degrade threshold value, a log warning message, syslog event and SNMP trap with the message ‟CRC errors in excess of the configured degrade threshold <M>*10e-<N> Set” is raised.

  • If the CRC error rate increases further and exceeds configured the signal fail threshold value, an alarm log message, syslog event and SNMP trap should be raised, and the port should be brought operationally down.

When the condition is cleared, a SNMP trap message to clear the event is sent out.