Hi,ZhiQiang,
I have the same confusion, coincidentally, for some metrics, samples can
be generated from both motification-agent and central-agent(or
compute-agent), such as image, floatingip, image.size..., these samples
from notification can effect the statistics of the meter, further, this
may broken alarm-evaluation base on the meters.
So,is it an issue?
For your question, think about using combination alarm or specifying
exclude_outliers in alarm.rule, just guess :)
? 2014/4/14 20:01, ZhiQiang Fan ??:
Hi, operators,
I have an environment running openstack havana which enables
ceilometer service, and I set notify_on_state_change =
vm_and_task_state in /etc/nova/nova.conf. So there are notifications
being sent when instances state and task status changed, the problem
that the end users may create alarm based on instance count, which
mea|ns |statistics on instance meter will be used, which leading to an
incorrect state of alarm, because there may be 5 vm but has 10
instances sample in 1 periods.
I think the cpu_util can replace the instance, but what about the vm
has multiple cpu?
So, it is really confused me how to check vm count through statistics api
Can anyone share your experience? Or is there some material can be read?
Thanks!
(sorry, the previous mail seems not display correctly, so I repost it
in reply.)
_______________________________________________
Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to : openstack@lists.openstack.org
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
_______________________________________________
Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to : openstack@lists.openstack.org
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
