I beg you pardon for the double-posting.
I paste here a couple of messages that I posted before in an old discussion (whose title referred to something different from the last post and, hence, in a less evident position) without receving any reply.

Our site is not properly functioning since almost a couple of weeks, and your help is essential to definitely solve the question!

Below a merge between the two messages.

Thank you

Andrea

----------------------------------------------------------------------------------

Some (not very encouraging) updates:
- we performed the upgrading;
- the system created a copy of users in the warehouse dbase, but only for a couple of users recently registered through the internet site;
- the number of users in the warehouse [lookuplist/people] is significantly smaller (about half) in comparison of that of the internet site [in Drupal: user management/users/list];
- after login, the data visualisable are those recorded by another user;
- in practice, the number ID of the warehouse is associated to a different user in comparison to that associated to the same number by the CMS;
- if a new user register himself, the system does associate to him the data already present in the dbase, previously recorded by a different user.
This "disallineation" caused, as you can imagine, heavy problems and we have to decide how to go ahead. Upgrading to 0.8.1 has to be done, but we are wondering if this (and expecially the easy login feature installation) had to be done in a system with an empty dbase!
I beg you pardon for a not-specific language, but I hope you understood our problem.
As usual, any help is wellcome!
Andrea

I paste below the comment of our web-server consultant, with the hope that it would be useful to solve our problem:
We think that the resulting error could be reduced after handling the two
tables for users (respectively inside Drupal and WareHouse environments )
in an opportune and unfortunately unknown manner. Yet not knowing the exact
data structure of the two environments we cannot know how to operate in an
effective and safe way.
Is there a way to correct the future insert operations and to transform
all previous data in a consistent manner ?
We think we have to follow an algorithm like this:
    1. An intermediate temporary table is created inside the Warehouse
    with 2 attributes :
       1. User id according to WH
       2. User id according to InstantIndicia
    2. Since (after EasyLogin ?) the next and unique authenticative table
    is the Users according to InstantIndicia and all observations are depending
    on that
        1. the new WH users table is rebuilt from old WH users and the
            transition table built at step 1
        2. the WH observations are rebuilt in the following way :
             1. data before the change of release : changed from old
             observation and table in step 1
             2. data after the change of release : they should be ok
Yet we do not know whether there are other depending tables and we do not
the data structure of the two databases (moreover they are different i.e.
Mysql and Postgresql).
Any suggestions ?