Ask Your Question
0

How does aodh/panko use data from Ceilometer?

asked 2018-12-06 05:22:29 -0600

pm gravatar image

In the current release Ceilometer API has been deprecated. As far as I understand , aodh and panko are extensions of ceilometer(all are treated as seperate projects) . But how do they manage to use data metrics from ceilometer ?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
1

answered 2018-12-06 05:55:16 -0600

I wouldn't call Aodh and Panko extensions of Ceilometer. Panko uses Ceilometer as a data source, but could use any other compatible telemetry tool. Aodh doesn't even interface with Ceilometer. Of course, they all belong to OpenStack's telemetry family, and developers probably cooperate closely.

Aodh manages and triggers alarms. For sample-based alarms, it gets the necessary data from a samples database, which is usually Gnocchi these days. For example, here is a snippet from /etc/ceilometer/pipeline.yaml on my cloud, which shows how Ceilometer is configured to send its samples to Gnocchi:

sinks:
    - name: meter_sink
      publishers:
          - gnocchi://?archive_policy=low&filter_project=gnocchi_swift

Panko sits right next to Gnocchi, but is optimized to store event data rather than samples. Ceilometer's event processing is normally configured in event_pipeline.yaml, and it is there that the event data destination is configured. It could be Panko, but my cloud (Devstack) uses Gnocchi for events.

To get data for event-based alarms, I suppose Aodh can use either Gnocchi or Panko.

edit flag offensive delete link more

Comments

Ohh things seems a clearer now thank you. But still i am wondering that you said Panko uses Ceilometer as a data source., but how does it manages to do that in the absence of ceilometer API. Does the data get stored in database for Panko to access? Or is there any other logic?

pm gravatar imagepm ( 2018-12-07 00:15:24 -0600 )edit
1

It's the opposite: Ceilometer uses Panko's and Gnocchi's API to send them the data it has acquired.

Bernd Bausch gravatar imageBernd Bausch ( 2018-12-07 01:47:03 -0600 )edit

Thank you.

pm gravatar imagepm ( 2018-12-07 02:44:54 -0600 )edit

But does it not make things a bit complicated for aodh and panko. Rather than sending the API request to Ceilometer for whatever data required, it is the ceilometer that has to push the data to the two. I am just being curious. and maybe wrong.

pm gravatar imagepm ( 2018-12-07 03:40:39 -0600 )edit

Why should pushing be more complicated than pulling?

The architecture is so that Ceilometer acquires and processes samples, then publishes them to various destinations, e.g. DBs, HTTP POSTs or UDP sockets. Before Gnocchi existed, it published data to a MongoDB by default.

Bernd Bausch gravatar imageBernd Bausch ( 2018-12-07 04:43:46 -0600 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2018-12-06 05:22:29 -0600

Seen: 20 times

Last updated: Dec 06