[katello-devel] APIs we need from Pulp (or candlepin)

[Jay Dobies]

On 10/25/2011 05:46 PM, Mike McCune wrote:
> The Pulp team is kicking off a new sprint next week and wants to know
> what APIs they can add to help speed up access to Pulp's data. Ideas
> include:
>
> * bulk status call to get repo sync status for a set of repos vs having
> to fetch it for each individual repo
> * bulk delete?
> * bulk create?
>
> Are there other things people have been wanting from Pulp to make
> Katello faster and more efficient?
>
> Mike

I'm not sure I like the idea of a bulk delete/create, at least not in 
the current architecture*

Delete is a potentially intensive command as it cleans up orphaned 
packages.

Create will potentially run into data validation issues, so if you made 
one bad ID out of 10, we'd need to do some sort of partial error support 
which we're not entirely decided on how we want to handle it.

Both are possible, but not critical IMO. When I asked about bulk 
operations I was more thinking data retrieval to be able to get the 
answer you want in a minimum number of server round trips.


* In the future I want to see these put through the tasking subsystem 
which would make their potentially resource-intensive aspects less of an 
issue.


-- 
Jay Dobies
RHCE# 805008743336126
Freenode: jdob @ #pulp
http://pulpproject.org | http://blog.pulpproject.org