Administrator Guide

Monitoring and Alerting Policy Suite Administrator’s Guide 47
53-1002933-02
Bottleneck detection integration with the MAPS dashboard
6
The following example clears only the dashboard summary data.
switch:admin> mapsdb --clear -summary
NOTE
When the dashboard is cleared, a RASLog message is generated. For more details on RASLog
messages in MAPS, refer to the Fabric OS Message Reference.
Bottleneck detection integration with the MAPS dashboard
The Fabric OS bottleneck daemon is responsible for detecting persistent bottlenecks and providing
notifications. Bottleneck monitoring is integrated with the MAPS dashboard enabling you to easily
see which ports are impacted by both persistent and transient bottlenecks. In the MAPS
dashboard, the Summary section includes bottleneck events detected by the bottleneck daemon,
and the Historical section displays entries both for those ports that have bottleneck time detected
by the daemon and for those ports that have “cred_zero” counters that are not zero.
NOTE
For the bottleneck daemon to produce notifications, the sub-second bottleneck monitoring
parameters must be correctly configured and the bottlenecks must be seen persistently on the ports.
Bottleneck events detected by the bottleneck daemon are shown in the Rules section of the
Summary view. However, even if the bottleneck daemon does not log a bottleneck event (due to
lack of persistence) the data shown in the Historical section can be used in the following ways:
• MAPS identifies the ports on which bottlenecks are seen and sorts them based on the number
of seconds that they exceeded the bottleneck threshold. This identifies the most strongly
affected ports, whether they are affected by persistent or transient bottlenecks.
• The cred_zero counter can also be used to detect bottlenecks. If the cred_zero counter
increases for a port but no bottleneck time is recorded, this indicates a potential transient
bottleneck on the port.
Additional information about bottleneck detection
The following information may help you to use MAPS as an aid to bottleneck detection.
• No existing bottleneck daemon logic and behaviors have been changed for Fabric OS 7.2.
• All bottleneck configurations must be made using bottleneckmon commands. Refer to the
“Bottleneck Detection” chapter in the Fabric OS Administrator’s Guide for specific command
details.
• The MAPS dashboard is used only for logging bottleneck latency events. Congestion bottleneck
events are not logged on the MAPS dashboard. Latency events include:
• Latency bottleneck on any port.
• Timeouts occurring on any 16 Gbps-capable Fibre Channel platform port.
• Stuck Virtual Channel on any port.
• The MAPS dashboard will continue to log events whether RASLogs are set to on or off in the
bottleneck configuration.