User story #202
closed
Store inventory revision in RDBMS
Added by François ARMAND about 14 years ago.
Updated about 9 years ago.
Category:
Web - Nodes & inventories
Description
For now, LDIF's inventory revisions are stored in the FS, in a at most naive way.
Store them in PostgreSQL
Of course, we will have to be a little smart about how to store them.
Most of the time, new revision of an inventory will be >90% the same the previous one.
As we inherently deals with trees, that make us think to Zippers (i.e http://scienceblogs.com/goodmath/2010/01/zippers_making_functional_upda.php)
At least, we should tar ldif (I believe that PostreSQL is able to make that automatically), and to not copy diff when not updated (so, two table: one [server uuid ; datetime ; ldif id], and one other [ldif id, ldif content] so that several (server uuid, datetime) are allowed to point to the same ldif content)
- Project changed from 2 to 20
- Target version changed from 24 to Ideas (not version specific)
Actually, we don't store inventory revision at all nowadays, and if we were starting to do so, we certainly won't do it in LDIF, nor in the database.
So closing that one as "you were relevant 5 years ago, not today"
- Status changed from New to Rejected
- Project changed from 20 to Rudder
- Category set to Web - Nodes & inventories
Also available in: Atom
PDF
Updated by 
Updated by 
Updated by