[openstack-dev] [Ceilometer]: Instance creation and deletion metrics in ceilometer !

gordon chung gord at live.ca
Wed Nov 2 14:01:46 UTC 2016

On 02/11/16 01:39 AM, Adrian Turjak wrote:
> Been vaguely following this thread and I have a question.
> Just to confirm, as I haven't touched ceilometer code in ages, the
> instance metric still exists? Or at least something like it?

it sort of exists currently. we don't build it from notifications but 
the pollster still generates it. (but it will be dropped unless people 
tell us otherwise.)

> We're currently using ceilometer as the data collection for our billing,
> and the instance metric is our primary way of billing for compute mainly
> because it tells us which instances exist at a given point in time. We
> further then use the metadata of that metric to get instance state
> (building, active, shutdown, terminating, etc) to determine if we should
> bill it.

i'm curious, what is the query you are using to get this information? 
the state information still is available via events which in the case of 
gnocchi, it uses to store resource state alongside metrics

> With the changes to ceilometer and the move to gnocchi I know we will
> need to rebuild how we handle billing data as we upgrade, but what I'm
> worried about is if gnocchi+ceilometer have some equivalent to the
> instance metric that will supply us with the same data, or do we now
> need to do our own notification monitoring...

i believe all the data is still available in some form. regarding 
gnocchi specifically, it takes sample data to capture measurements for a 
metric and takes event data to capture state (and possibly other metadata)

> Basically what I need is time series data of, this instance was in this
> state from this period to that period, and if I query for a range I get
> the ranges or changes in that time series. Is something like that
> present, or if not would I be able to make something like that in
> gnocchi? That way I can then query for a time range and know what state
> changes occurred for a given instance.

gnocchi captures the state of a resource and it's history. this is 
accessible by looking at resource history. i'm not entirely sure if that 
handles your case, may you could provide the queries you use and we 
could figure out equivalent gnocchi queries. i built a ceilometer vs 
gnocchi usage deck[1] that may help but it's more focused on metrics 
rather than resource history.

[1] http://www.slideshare.net/GordonChung/ceilometer-to-gnocchi


More information about the OpenStack-dev mailing list