Well, I totally forgot about that one…that truly simplified this case and I did 
not have to re-invent the wheel again! I thank thee for that!...:)

Even though the attributes left to correlate with are a bit ambiguous and we do 
have data quality problems here…



From: Francesco Chicchiriccò [mailto:ilgro...@apache.org]
Sent: torstai 15. kesäkuuta 2017 18.03
To: user@syncope.apache.org
Subject: Re: Pull users from two primary sources with different UID:s

Hi Mikael,
if I understand your environment, the problem can be seen as the general 
situation where there are two distinct external resources, each with a 
different unique id, matching two distinct attributes in Syncope.

In such situations, at least one of the resources  bears one or more attributes 
that can be used to match users on Syncope, even if the key is not (yet) 
available (say Firstname + Surname, even though I know I am over-simplifying).

The tool to leverage here are the Pull Policies [1], in particular the Pull 
Correlation Rules [2].


[1] https://syncope.apache.org/docs/reference-guide.html#policies-pull
[2] https://syncope.apache.org/docs/reference-guide.html#pull-correlation-rules

On 14/06/2017 16:05, Mikael Ekblom wrote:

We have a small challenge here that includes two primary pull sources, where 
the UID for these sources are not the same or cannot be set to the same value. 
The only common UID(National Identification Number) that could be used as a 
kind of primary key for users pulled from both sources (that would make this an 
easy case) can change to my surprise when it comes to foreign exchange 
students.This is when they get Finnish versions of the NIN.  It will not be 
possible to set any previousUID column to both sources either (for payroll 
particularly) and I would like to avoid having to build any separate repository 
only due to this case.

So, what I do during provision and within beforeProvision to be precise, is 
that I check if I get any existing match for the pulled user based on the NIN. 
If I do, I like to create a new syndelta with the information from the existing 
user together with the fields that should be added. If no match is found, then 
the provisioning should continue as before.

What I do is to create a new syncdelta with the UID changed to the existing UID 
for the account matching, set the connector object with the required fields, 
create an AnyTO object and an anyPatch object and send this to beforeupdate 
together with the new delta.

Everything is fine so far.

Within beforeUpdate, I can see that I get all the required fields for that 
connector object. AbstractAnyDataBinder though claims that I have a required 
attribute missing: eduPersonUniqueId, that I can see that is present within 
both objects (AnyTO and AnyPatch ) sent to the beforeUpdate and it is also 
available as a field within the connector object for the sync delta.

What am I missing? Should I recreate the syncdelta object again within 
beforeUpdate together with the AnyTO and AnyPatch objects too, to be sure? What 
is required for beforeUpdate to function properly? Is it even possible to 
revert to beforeUpdate from say beforeProvision if a possible match is found? 
It should I think. It is a bit troublesome to check for yourself, when the 
API-documentation is not that detailed…☺




Francesco Chicchiriccò

Tirasa - Open Source Excellence


Member at The Apache Software Foundation

Syncope, Cocoon, Olingo, CXF, OpenJPA, PonyMail


Reply via email to