What is this alert?

The metrics collector extension checks CPU utilization on the master host every 15 seconds. If the percentage of CPU in use is higher than the percentage specified in the alert rule for the number of minutes specified in the rule, Command Center raises an alert.

The Greenplum Database master uses the greatest amount of CPU at the start of a query, while planning the query, and at the end of the query, while gathering results from the segments. For a large result set it is normal to see a spike in the query's CPU usage on the master during the gather operation. With many queries running concurrently, the CPU spikes should even out.

What to do

Begin by viewing the Command Center Query Monitor and Query Details views to see if there are one or more very large queries nearing completion, or if the high usage can be explained by unusual workloads or heavy query activity.

If the query monitor offers no explanation for high CPU usage, you should investigate master host processes using Linux utilities such as ps and top to identify processes consuming the CPU. If the process ID of an errant process is a postgres process, you can query the pg_stat_activity system table to find the query and, if needed, use the pg_terminate_backend() function to terminate the query.

See also

Investigating a Performance Problem

check-circle-line exclamation-circle-line close-line
Scroll to top icon