[openstack-dev] [nova] Ocata blueprint burndown chart

Matt Riedemann mriedemos at gmail.com
Fri Feb 3 22:37:06 UTC 2017

For those that like charts, I was keeping track of our progress with 
approving and then completing blueprints targeted to Ocata over the 
release. It's pretty unscientific. I basically had a spreadsheet with a 
cell per day and I'd just plug in numbers from launchpad based on the 
total number of targeted blueprints, the number that were approved (i.e. 
spec was merged) and when they were completed. I also had milestone 
markers in there for reference.

The resulting chart is here [1].

Some observations:

- There is an obvious correlation between ramping up targeted 
blueprint/specs and approving them toward the spec freeze with a jump at 
the end. That shouldn't be a surprise.

- We started getting more approved blueprints merged/completed after the 
spec freeze, but the rate on that was still very low until about 
January. The spec freeze was on November 17th so it's not terribly 
surprising to me that even after that, completion rates were slow until 
January since we were hitting the end of the year when a lot of people 
were taking time off for holidays.

- We had a blueprint code review sprint on January 11th at which point 
completed blueprints spiked up a bit. After that it was a dash to 
feature freeze.

- Just a general note that the targeted/approved lines drop off and meet 
the completed line at the end due to me deferring blueprints that didn't 
make Ocata.

- At the peak we had 78 targeted blueprints and 70 approved blueprints. 
That's less than Newton but Ocata was also a shorter cycle and we had 
fewer resources on hand for reviews in this release. Comparing final 
numbers to Newton, we had 47 completed or partially completed blueprints 
in Ocata compared to 64 in Newton (at one point we had 100 approved 
blueprints for Newton too I think). So rate-wise those releases were 
about the same (67% completed in Ocata vs 64% completed in Newton). And 
really, that's surprising to me personally given we had pretty much 3 
less major nova core reviewers in Ocata (danpb, alaski and sdague).

I haven't fully digested this all yet, or really had too much time to 
reflect on it in detail because this has been a hectic past few weeks.

We'll do a retrospective for Ocata like we did for Newton and go through 
that at the PTG.

I just wanted to get this chart out as it's something I've been tracking 
for the past 5 months and I'm happy to finally post it.




Matt Riedemann

More information about the OpenStack-dev mailing list