10g replication

Hi friends,

I was able to set-up my MATERIALIZED VIEW replication between my two remote databases. My base/master table contains around 1 million+ rows.
I have to pre-create it to the materialized view site because it will be too long
to refresh it (complete) over WAN. What I did was export/ftp/import it to
the target site, which is faster to do. Then using the OEM 10gR2 client I set it
up the MASTER SITE/MATERIALIZED VIEW SITE and I was successful.
I was monitoring the the replication activity and it was fine for the first week,
I counted both tables everyday/hr/min and they have the same count.
Then I left it alone for 1 month without checking. When I come back.
I saw that the count were not the same anymore. and I checked the mlog$
table and I see that there are changes there not refreshed to the materialized
view site. I also check the REFRESH GROUP STATUS but it was NORMAL.
Unlike like before it turn to BROKEN with it can not refresh the target view.
Why dees it behave like this? It seems that there is no problem with the
connection because of the status NORMAL. but it was not doing the replication job. I tried to refresh it manually by clicking the refresh button.
But It was taking a long time and got hanged. How do I check and trace the
cause of the problem?

Thanks a lot