Open Stack

On 09/22/2018 11:15 AM, Matt Riedemann wrote:
> On 9/21/2018 4:19 PM, Ben Nemec wrote:
>>> * The only two projects that I'm aware of with patches up at this 
>>> point are monasca [2] and designate [3]. The monasca one is tricky 
>>> because as I've found going through release notes for some projects, 
>>> they don't really have any major upgrade impacts so writing checks is 
>>> not obvious. I don't have a great solution here. What monasca has 
>>> done is add the framework with a noop check. If others are in the 
>>> same situation, I'd like to hear your thoughts on what you think 
>>> makes sense here. The alternative is these projects opt out of the 
>>> goal for Stein and just add the check code later when it makes sense 
>>> (but people might forget or not care to do that later if it's not a 
>>> goal).
>>
>> My inclination is for the command to exist with a noop check, the main 
>> reason being that if we create it for everyone this cycle then the 
>> deployment tools can implement calls to the status commands all at 
>> once. If we wait until checks are needed then someone has to not only 
>> implement it in the service but also remember to go update all of the 
>> deployment tools. Implementing a noop check should be pretty trivial 
>> with the library so it isn't a huge imposition.
> 
> Yeah, I agree, and I've left comments on the patch to give some ideas on 
> how to write the noop check with a description that explains it's an 
> initial check but doesn't really do anything. The alternative would be 
> to dump the table header for the results but then not have any rows, 
> which could be more confusing.
> 

+1 to "this page intentionally left blank", hopefully without the 
logical contradiction those pages always create. ;-)