How to solve problems with synchronization of distributed databases 1C (URIB)?

Question

Data synchronization

There is 1C 7.7, one central base and 2 peripheral ones. Periodically, there are problems with synchronization between them - for example, 2 databases see some kind of document, but the third does not. Either there is a document, but there have been no movements in the registers.
This was solved by reposting the documents and restoring the sequence. Although in theory 1C should control the integrity, etc. Our databases are synchronized every 5-10 minutes, the amount of data transferred is small.

Now here is a new error - different bases see different balances in the Money register at different cash desks. And this, depending on the error, can be mistaken for a shortage at the cashier.

The question is: how to make sure that everything is fine and all three bases are identical? If this is not realistic, then how can you at least quickly track down problems?

Answer 1

It is better to ask this question at http://www.forum.mista.ru , on Habré it is not very common to write articles and answer questions on 1C.
Here is an example:
http://habrahabr.ru/post/83903/ , or here it is
http://habrahabr.ru/qa/21005/

Answer 2

Physically, how is the exchange implemented?
I am unfamiliar with the seven, but in the eight I would immediately sin on the synchronization mechanism or broken packets.
those. if we synchronize via OLE, then we look at the code for synchronization, if ftp then we check the packets for validity ...
carefully read the exchange logs. Sometimes there are interesting things there. For example, on one server, ms-xml regularly disappears somewhere ...