GigaSMART CPU Utilization Statistics

You can display CPU utilization statistics for GigaSMART. The statistics indicate the performance of GigaSMART, improve visibility, and help identify high load conditions.

Show commands display instantaneous CPU utilization as well as historical, providing trends for CPU utilization.

You can also configure a rising threshold, as a percentage, to indicate when high CPU utilization occurs. When the aggregate CPU utilization percentage exceeds the rising the threshold, an SNMP notification can be triggered.

This feature is supported on all products that support GigaSMART: GigaVUE‑HC1, GigaVUE‑HC2, and GigaVUE‑HC3.

The GigaSMART engine port (e port) numbers are e1 on nodes with one GigaSMART engine and e1 and e2 on nodes with two GigaSMART engines, for example: 10/1/e1 or 8/1/e1 and 8/1/e2.

Refer to the following sections for viewing statistics, configuring the threshold, and configuring a notification that can be sent when the threshold is exceeded:

■   Display GigaSMART CPU Utilization
■   Configure Threshold
■   Configure Threshold Crossing Notification

Use the show gsgroup stats all command to display the statistics on all GigaSMART groups on the node.

The statistics are displayed for 1 second, 1 minute, 5 minute, 10 minute, and 15 minute intervals. The 1 second interval displays the statistics for the previous second. The 1 minute, 5 minute, 10 minute, and 15 minute intervals display statistics containing history.

Statistics are displayed in an aggregate form. For example, if there are two GigaSMARTe ports: e1 and e2, there will be two aggregates. One aggregate will be for e1, the other will be for e2. The term aggregate refers to aggregation across all packet processing cores (up to 31) in the CPU. It does not refer to an aggregate across CPUs.

The statistics are as follows:

■   Useful Time—Amount of time during which the CPU is processing packets, in milliseconds (ms) or seconds (s).
■   Idle Time—Amount of time during which the CPU is not processing packets, for example, when it is busy looping, in milliseconds (ms) or seconds (s).
■   In Packets (pkts/s)—Number of packets per second coming into the CPU. For the 1 second interval, In Packets is the actual number of incoming packets for that second. For the 1 minute, 5 minute, 10 minute, and 15 minute intervals, In Packets is an average number of incoming packets per second.
■   Packets Drop (pkts/s)—Number of packets per second dropped by the CPU. For the 1 second interval, Packets Drop is the actual number of dropped packets for that second. For the 1 minute, 5 minute, 10 minute, and 15 minute intervals, Packets Drop is an average number of dropped packets per second.
■   Packets Recv Error—Number of received packets per second with errors. For the 1 second interval, Packets Recv Error is the actual number of errored packets for that second. For the 1 minute, 5 minute, 10 minute, and 15 minute intervals, Packets Recv Error is an average number of errored packets per second.
■   CPU Usage %—Percentage of time during which the CPU is processing packets. CPU Usage % plus CPU Idle % equals 100.
■   CPU Idle %—Percentage of time during which the CPU is not processing packets. CPU Idle % plus CPU Usage % equals 100.

Note:  When the node is restarted, the 1 minute, 5 minute, 10 minute, and 15 minute statistics will not be exactly for 1 minute, 5 minute, 10 minute, and 15 minute intervals, until the full interval has elapsed and the history is available.

An upper threshold (rising) can be configured. When the aggregate value of the CPU utilization on the GigaSMART engine exceeds the threshold, an SNMP notification can be triggered. Refer to Configure Threshold Crossing Notification.

When the aggregate value of the CPU utilization exceeds the upper (rising) threshold, a message is logged, and optionally, an SNMP notification is sent to all configured destinations.

When enabled, the SNMP notification is sent when the rising threshold is exceeded in any 5-second interval over which the CPU utilization is averaged.

Note:  Once the rising threshold is exceeded for 5 seconds, the SNMP notification is generated. However, if the CPU utilization falls below the upper threshold but does not remain below that threshold continuously for 5 seconds, a new notification is not generated when the upper threshold is exceeded again. A new notification is generated only when the CPU utilization falls below the threshold, stays below the threshold continuously for 5 seconds, exceeds the threshold again, and stays above it for 5 seconds.