OpenStack Ceilometer capacity utilization metrics setting


#1

Hello,

when there is running metrics collection at OpenStack provider, Ceilometer is over-utilized and we are getting error responses. Using current CFME (5.6.1.2).

What are the options for CFME metrics collections settings?
:ems_metrics_collector_worker:
:defaults:
:count: 2
:nice_delta: 3
:poll_method: :escalate
:ems_metrics_collector_worker_openstack: {}
:ems_metrics_collector_worker_openstack_infra: {}
:ems_metrics_collector_worker_openstack_network: {}

Is it possible to modify frequency of utilization data collection?

THX in advance.


#2

Hi @vmi,

There are two different settings that are used to control how often metrics are collected:

  1. The Schedule Worker settings
    https://github.com/ManageIQ/manageiq/blob/46277eefa4ea925a2607628c7815fbc5a8f214c0/config/settings.yml#L1412

The Schedule Worker is set to check every three minutes for any targets that need metrics collected. But, this doesn’t mean MIQ is querying against Ceilometer for metrics every three minutes. Only that MIQ is looking for anything in the MIQ database that hasn’t had metrics collected “in a while”.

  1. The Capture Threshold:
    https://github.com/ManageIQ/manageiq/blob/46277eefa4ea925a2607628c7815fbc5a8f214c0/config/settings.yml#L961-L966

The Capture Threshold settings define the “in a while” that the Schedule Worker is looking for.

So, putting these two things together, the Schedule Worker is checking every three minutes for any targets that last captured their metrics later than their capture threshold.

Hope that helps.