[GnomeMeeting-list] Problems with adressbook

[Michael Neumeier <dennismail gmx net>]

I have a little problem with the adressbook of gnomemeeting 1.0.1 under
SuSE 9.1. I start my system, which is connected to a Netgear DG824B
ADSL-Modem-Router. The Firewall of the router is configured to let
through all H323-Packages to a specific PC in my LAN. 

When I open the adressbook of gnomemeeting and search all contacts on
ils.seconix.com, it shows them. After a few minutes, I do exactly the
same and then I get the "contact-message" that says: "If you registered
and see this message, it means that you are unreachable from the
outside" (and it shows 3 users of 4 found!). If I search for a specific
user, e.g. all users whose first name is "michael", the adressbook shows
me all users who fit this search. 
This would mean that it is not a problem of the firewall of the router
(if I�m right), because otherwise it wouldn�t find users in specific
search.  

The next thing to look at is the log of the firewall. I configured the
router to log all things happening, no matter if the rule for
h323-packages fits or doesn�t fit. When I made sure that a search for
all users is not working anymore, I cleared the whole log and seached
again. Now, the log tells me that the inbound rule does not fit (and
then drops the package!). That leads us to the rule itself:
It covers the rule for the H.323-service with tcp-packages on port 1720.
I did set the firewall to ALLOW ALWAYS for these packages, just to be
sure that these packages can always pass. They are routed to a specific
PC in my LAN from the firewall. The last option I can set is "WAN Users"
which is explained this way: "These settings determine which packets are
covered by the rule, based on their source (WAN) IP address. Select the
desired option", which I did set to "any", this means that all IP
adresses are covered by this rule. 
Back to the log again: The source of the packages which does not fit are
coming from different Source-IPs. I know that there may be packages who
are not h.323-packages and can be seen as blocked in the log, but I
think there are also h.323-packages among the blocked ones.

Does anybody has an explanation for this strange behaviour?

Yours, Mike