[cloudkitty] Antelope PTG summary

Rafael Weingärtner rafaelweingartner at gmail.com
Tue Oct 25 11:00:18 UTC 2022

Dear all,

The CloudKitty project had a great PTG meeting. Thanks to the
OpenInfra Foundation for holding it. Also, thanks to all who have dedicated
precious time to attend.

We started the meeting by revisiting the latest release and discussing the
review format we have adopted in the past two years. Looking at the result
of our process, we decided to maintain it, as it seems to bring results and
concentrate our efforts on getting CloudKitty better and evolving
throughout time.

While reviewing the past release, we discussed the situation with patch [1]
(Allow rating rules that have 12 digits in the integer part of the number).
This patch is a bug fix, and we discussed the idea of backporting it.
However, as it is a patch that affects the database schema, it can be
troublesome to backport it. Therefore, before doing any work, we will
consult the mailing list to gather others opinions on this matter. Users
could upgrade CloudKitty to address this issue, if they need to, instead of
applying a patch. Pierre volunteered to execute this "inquiry" with the
CloudKitty user base in the mailing list.

While discussing the ElasticSearch (ES) improvements in CloudKitty to
support ES 7, we raised the concern that Kolla-ansible is moving from
ElasticSearch to OpenSearch. Therefore, we need to evaluate if this can
have an impact in CloudKitty. Most certainly it will impact people using ES
as the backend. The goal is to revisit this discussion in our next
bi-weekly meeting, and then communicate for users and community (Topics for
meeting of 31/10/2022 [2]).

Furthermore, as the current CloudKitty team is a bit small, we have decided
to focus on the following topics for the Antelope cycle.

   - Multiple rating types for the same metric -- One metric being used to
   generate different rates
   - Different time frame group by options -- provide group by options to
   group by day of the year, week of the year, month, and year.
   - Improve some timeouts handling when using Gnocchi backends. This one
   needs some checking with Zigo, who was the one that reported this
   situation. Zigo, I added you to this e-mail, can you reply back with more
   details regarding this situation?
   - A regression API that takes the resource ID Or project ID or scope ID,
   and executes a regression (prediction) based on the collected data so far
   for the given resources/elements/scope.
   - Add CloudKitty API reference docs (
   to https://docs.openstack.org/api-ref/

We also briefly discussed the situation with Monasca, which has already
been deprecated from Kolla-ansible, and might be losing some traction and
support. Therefore, we (as a community), might need to start thinking about
deprecating it, and removing some non-tested code from our code base. We do
understand that deprecating and removing support is complicated for users,
but nobody is maintaining and testing that integration. Therefore, it tends
to get broken and unstable over time. However, if somebody wants to
volunteer to maintain and improve it, you are welcome :)

In summary that is the outcome of our small, but efficient PTG meeting.
Let's keep improving and making CloudKitty better. If I missed something,
please do not hesitate to reply back. Also, if somebody needs any help, you
can always ping me :)

See you guys. Have a nice week!

[1] https://review.opendev.org/c/openstack/cloudkitty/+/837200
[2] https://etherpad.opendev.org/p/cloudkitty-meeting-topics

Rafael Weingärtner
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.openstack.org/pipermail/openstack-discuss/attachments/20221025/9c8cb2d6/attachment.htm>

More information about the openstack-discuss mailing list