Some delicate considerations need to be taken into account for the synchronisation process in case a user has reconfigured how two storages synchronise collections with each other.
This task includes implementing such protections and some explanation on how they work and the dangerous situations which they prevent.
This is milestone 24
Given how
mapping_uid
s are fetched from the status database, I think this is mostly covered.I mostly need to audit this. When reconfiguring a collection mapping, the status db will have data for the old mapping. None of this data should be used in this case.
Does it make sense to also flush data for the old mapping?
I might want to add a
fingerprint
function to mappings, which is like a platform-independent hash to compare if they have changed or not.
Hugo Osvaldo Barrera referenced this ticket in commit 92f2c19.