21-7
Ethernet Card Software Feature and Configuration Guide, R7.2
January 2009
Chapter 21 Configuring RMON
Threshold and Triggered Actions
The user can also configure the CRC-ALARM to trigger a l ink state down on the port and to wrap an
Cisco proprietary RPR . By default, the CRC-ALARM is disabled. When the alarm is configured, the
link down and wrap actions are still disabled by default. This feature is also supported on the ML-Series
card Ethernet ports.
Threshold and Triggered Actions
The configureable threshold is not set with a BER, since variable frame lengths and varying percentages
of bandwidth can impair the usefulness of this measure. I nstead, the users configure a more relevant
measure using CRC error rate as a percentage of the traffic. The available triggering thresholds are:
• 10e-2 or 1% traffic (1 CRC error 100 packets)
• 10e-3 or 0.1% traffic (1 CRC error in 1000 packets) (default)
• 10e-4 or 0.01% traffic (1 CRC error in 10000 packets)
The default threshold is a CRC error rate of 0.1% of the traffic. For voice and video traffic, an error rate
of 1% is typically a critical issue and 0.1% is a major issue. Voice and video needs to trigger a wrap if
the error rate is higher than 0.1% (1 error every 1000 packets). For normal data traffic, an error rate of
10% traffic is a critical issue, requiring an immediate fix, and 1% traffic is a minor issue.
The following actions occur after the detection of excessive CRC errors:
1. The Cisco proprietary RPR wraps if this option is configured.
2. The link shuts down if this option is configured.
3. If the link shuts down, a path defect indication (PDI) is sent to the far-end ML-Series card port. This
ensures the remote end wraps.
4. A CRC-ALARM is raised against the local end ML-Series card port. (If the remote end is also
receiving excessive CRC errors, a CRC-ALARM is also raised against the far end ML-Series card
port.
SONET/GFP Suppression of CRC-ALARM
This detection of excessive CRC errors is independent of SONET/GFP defects. A problem may have the
potential to trigger both the SONET/GFP defects and the CRC-ALARM. In this scenario, the
SONET/GFP defect will trigger before the CRC-WRAP alarm because CRC error threshold detection is
a slower process. If the SONET/GFP defect causes the link to go down, this link-down happens before
the CRC-ALARM is detected, and it suppresses the CRC-ALARM. If the SONET/GFP defect that
causes CRC-ALARM is not a link-down trigger and the CRC-ALARM is configured to take the link
down, the CRC-ALARM will report and trigger the link down.
Clearing of CRC-ALARM
When the trigger action is disabled (default), the CRC-ALARM automatically clears when the error rate
falls below the threshold for a significant time period.
When the trigger action is enabled, a CRC-ALARM requires a manual clear from the user. This is
required because the wrap or link down caused by the alarm blocks both traffic and the CRC errors in
the traffic from the port. So with no CRC errors, an automatic clear would occur even though the
underlying problem, such as dirty fiber or a defective ML-Series card, still exists. Interface flapping can
occur in this situation.