Open topic with navigation
Each broker watches all domain events and reported metrics and fires alarms based on its view of the domain.
Statistical alarms based on metrics do not retrieve the metrics data from the
nuodb_system database. They are based on the last known values reported periodically either by the operating system or by NuoDB database processes. Database processes (TEs or SMs) as well as brokers report simple statistics/metrics every
n seconds. Each broker then aggregates statistics from all hosts in a given database into a database aggregate. The domain then aggregates all statistics for all databases into a single domain aggregate. Included in the domain aggregate are aggregates of all OS metrics provided by the agents. See Using JMX Beans to Obtain Metrics.
Domain events are reported periodically, every 30 seconds.
From most verbose logging to least verbose logging, the alarm severity levels are:
An alarm action is one of the following:
NotifyEventsends the alarm to any management client.
LogOnlyonly logs the alarm in the broker's agent log file. See Agent Log Description.