[openstack-dev] [Nova] Frustrations with review wait times (was: Requesting feedback on review 35759)

Russell Bryant rbryant at redhat.com
Tue Aug 27 14:31:32 UTC 2013


On 08/27/2013 03:35 AM, Gary Kotton wrote:
> 
>> -----Original Message-----
>> From: Wang, Shane [mailto:shane.wang at intel.com]
>> Sent: Tuesday, August 27, 2013 6:31 AM
>> To: OpenStack Development Mailing List
>> Subject: [openstack-dev] [Nova] Requesting feedback on review 35759
>>
>> Hi,
>>
>> We submitted the patches for bp
>> https://blueprints.launchpad.net/nova/+spec/utilization-aware-scheduling
>> 1+ month ago.
>> The first patch is to add a column to save metrics collected by plugins -
>> https://review.openstack.org/#/c/35759/.
>> Is there anyone who is interested in that, would it be possible to get some
>> reviews for that?
> 
>  [Gary Kotton]  I have taken a look. I too have patches in from over a month ago and understand the frustration... 

I'm sorry to hear that you guys are frustrated.

This probably doesn't make you feel better, but I do regularly track how
we're doing in this area.  We're currently a bit further behind than I
like us to be.  This isn't a big surprise though, as we are currently
right in the middle of the peak of review activity for the entire
development cycle (leading up to the feature freeze).  Otherwise, we
tend to stay ahead of our goals.

The goals we have set for Nova are to stay below the following by the
end of the work week, since the weekend skews things a bit:

    * < 4 days since the latest revision
    * < 7 days since the latest revision without a -1/-2

Here are the current stats for Nova:

    http://russellbryant.net/openstack-stats/nova-openreviews.html

and for the archive:

> Projects: [u'nova']
--> Total Open Reviews: 377
--> Waiting on Submitter: 102
--> Waiting on Reviewer: 275
--> Stats since the latest revision:
----> Average wait time: 5 days, 4 hours, 38 minutes
----> Median wait time: 4 days, 2 hours, 44 minutes
--> Stats since the first revision:
----> Average wait time: 18 days, 9 hours, 49 minutes
----> Median wait time: 13 days, 3 hours, 35 minutes
--> Stats since the last revision without -1 or -2 (ignoring jenkins):
----> Average wait time: 8 days, 16 hours, 32 minutes
----> Median wait time: 5 days, 23 hours, 36 minutes


Despite being a bit behind, Nova is still ahead of the average review
wait times across all of OpenStack (but above the median).

    http://russellbryant.net/openstack-stats/all-openreviews.html

> Projects: [u'ceilometer', u'glance', u'heat', u'neutron', u'infra',
u'grenade', u'keystone', u'ironic', u'oslo', u'docs', u'devstack',
u'cinder', u'nova', u'tempest', u'swift', u'horizon', u'tripleo', u'trove']
--> Total Open Reviews: 1043
--> Waiting on Submitter: 335
--> Waiting on Reviewer: 708
--> Stats since the latest revision:
----> Average wait time: 5 days, 23 hours, 39 minutes
----> Median wait time: 3 days, 23 hours, 13 minutes
--> Stats since the first revision:
----> Average wait time: 18 days, 7 hours, 41 minutes
----> Median wait time: 11 days, 2 hours, 13 minutes
--> Stats since the last revision without -1 or -2 (ignoring jenkins):
----> Average wait time: 9 days, 10 hours, 14 minutes
----> Median wait time: 5 days, 13 hours, 7 minutes


In general, how do we improve these stats?

1) Make sure we have a dedicated review team.

This is something I spend time on regularly.  I try to encourage those
on the team to regularly participate in reviews.  For those that aren't,
they generally get removed from the team.  I'm also always on the
lookout for people we should be adding.

    https://wiki.openstack.org/wiki/Nova/CoreTeam

2) Review prioritization

Reviewers choose which reviews to go after based on all kinds of things
and some get more or quicker attention than others.  That will always be
the case.  If there's a lot more interest in X than Y, then X deserves
more attention, honestly.

However, many of us could still do a better job prioritizing reviews.
You can find some tips on that here:

    https://wiki.openstack.org/wiki/Nova/CoreTeam#Review_Prioritization
    https://wiki.openstack.org/wiki/ReviewWorkflowTips

3) Something much more drastic

At some point we may grow to the point where we have to adopt a new
model to scale further.  I don't think we're too far off.  However, I
think we could stand to grow a dedicated nova-core a bit larger before
we have to take that step.  We have 19 people, and not all of those are
active right now (meaning the team will likely shrink a bit soon).

-- 
Russell Bryant



More information about the OpenStack-dev mailing list