Open Stack

I personally like Openstack Health UI/UX. I use it a lot a an initial
source of information.
Thanks to the OH team for all the great work on this!

And +1 to the new "feedback" link on the top and bottom of every page :)

A few thoughts on my experience with OH.

It can take a few clicks in OH to get to the view one's looking for - and
some views are slow to load.
It is much better now that the URL in the browser is always updated to
reflect the current view - very useful for bookmarking and sharing.
One limitation to that is the sorting of columns is not reflected in the
URL [0].

If I want to work on getting a new job in the gate, I'll create it as a
non-voting check or experimental job first, and there's no data for them in
OH yet.

If I want to check why an existing job is failing (e.g a periodic job [1])
I can see which tests are failing and how often, which is a great starting
point for my work, but if I click on any of the tests the job filtering is
lost. Seeing that a test is successful in some jobs and it fails in others
is good data, but I need to be able to see where are the failures coming
from. Ideas on how to implement that are:
- have the ability of applying multiple metadata based filters to a view
- have a list of the available metadata, and view the distribution of value
on each metadata field for the current view - a la ELK. So for instance if
I select so see only failures in  a graph, show the distribution only on
failures. If I click on a data point, show the distribution only for that
point. That way I can see where certain deviations are coming from
- get the list of logs associate to a certain point in the graph for
further analysis

Injecting the metadata in the job name can help overcome filtering
abilities a bit, but we cannot inject metadata in the test names, so at
test level the context is always lost.

I use OpenStack Health downstream as well, to track results from a
combination of test systems (CI and more), and the fact that both the DB
schema and the dashboard are metadata agnostic is very useful - I have
extra metadata which is inject in certain jobs, like test rack name and
product version. But again I need the ability to select on multiple
 metadata keys at once, so I can produce test views / reports for specific
software versions and test environments.

andrea

[0] https://bugs.launchpad.net/openstack-health/+bug/1577420
[1]
http://status.openstack.org/openstack-health/#/job/periodic-tempest-dsvm-neutron-full-test-accounts-master?groupKey=project&resolutionKey=hour&end=2016-05-02T14:31:26.423Z




On Fri, Apr 29, 2016 at 8:15 PM Masayuki Igawa <masayuki.igawa at gmail.com>
wrote:

> Hi,
>
> Now, we are making OpenStack-Health[1] which is a dashboard for
> visualizing test results of OpenStack CI jobs. This is heavily under
> development mode but works, you can use it now.
>
> We'd like to get your feedback to make it better UI/UX. So, if you
> have any comments or feedbacks, please feel free to file it as a
> bug[2] and/or submit patches[3].
>
> This is a kind of advertisements, however, I think openstack-health
> could be useful for all projects through knowing the development
> status of the OpenStack projects.
>
>
> [1] http://status.openstack.org/openstack-health/
> [2] https://bugs.launchpad.net/openstack-health
> [3] http://git.openstack.org/cgit/openstack/openstack-health
>
> Best Regards,
> -- Masayuki Igawa
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160502/b1850589/attachment.html>