[openstack-dev] [Ceilometer] Doubt about data recollection

Alberto Gutiérrez Torre alberto.gutierrez at bsc.es
Mon Apr 4 15:28:06 UTC 2016


In the end the RabbitMQ approach would not suffice our requirements as
in our installation we have RabbitMQ in the controller.

I'm trying to do it with the pipeline approach but trying to maintain
the API working. I did the following with cpu_util:

    - name: cpu_sink
          - name: "rate_of_change"
                    name: "cpu_util"
                    unit: "%"
                    type: "gauge"
                    scale: "100.0 / (10**9 *
(resource_metadata.cpu_number or 1))"
          - notifier://
          - udp://

I want to publish it in both channels but, when this is done, neither
work. If I try sending only data thought UDP it also refuses to work.
Any clues on this? I'm testing it with netcat listening to that port.

Also, is it possible to send an AMQP message to other IP that is not the
one of the controller? In the documentation only makes reference to the
options available for this kind of message, there is no place to specify
the host.

Thanks for your support,

On 15/03/16 16:48, gordon chung wrote:
> On 15/03/2016 11:01 AM, Alberto Gutiérrez Torre wrote:
>> What we would like to have is access to the data available in Ceilometer
>> API but by local access. For instance, CPU metrics produced at a given
>> node will be consumed in that same node. Probably we will not need all
>> the available data we're yet finding which variables are of our
>> interest. To start with, CPU usage and memory usage are mandatory.
>> I believe that what we are searching is to access directly to compute
>> agent data. Is this possible, for example, subscribing to  the RabbitMQ
>> queue where the data is being published by the agent? Is there any other
>> way to do so?
> that's exactly one way to do it. the (compute) polling agents build a 
> set of metrics for each poll and publishes them to message queue. the 
> default consumer of these messages is the notification agent which 
> processes the data further (if necessary) and redirects it to a storage 
> or alternative target.
> you can configure your compute agent publish to multiple queues by 
> setting notification_topic = notifications, <custom_topic>. or you can 
> just disable notification agent and all other non compute agent services 
> if all you want is the data
> alternatively, you can use all the services and have the notification 
> agent publish the data to a specific target. to do this, you'll need to 
> edit pipeline.yaml file to send just compute agent metrics to where you 
> want.
> cheers,

WARNING / LEGAL TEXT: This message is intended only for the use of the
individual or entity to which it is addressed and may contain
information which is privileged, confidential, proprietary, or exempt
from disclosure under applicable law. If you are not the intended
recipient or the person responsible for delivering the message to the
intended recipient, you are strictly prohibited from disclosing,
distributing, copying, or in any way using this message. If you have
received this communication in error, please notify the sender and
destroy and delete any copies you may have received.


More information about the OpenStack-dev mailing list