Hi, The behavior you've described is perfectly valid: gnugk does not care and should not care about routing decisions, all it does is seting of source address of the packets and then passes packets to linux routing subsystem to make decisions via which interface packet should be sent. By default linux routing is based only on packets destination address, when it is making routing decision it does not care about packet's source address (the on configured by setting "Home="). However recent kernels can be configured to take into account almost any ip header field including source address. More documentation regarding linux routing can be found at www.lartc.org. I'm not using openphone so I do not know answers for q4&q5 but for q6 I suppose that gnugk receives 2 GRQ's on 2 interfaces so this produces 4 replys. Julius On Thu, 2004-02-19 at 17:26, Kramer, R.D.J. wrote: > Dear all. > > Hmmm... it seems that I mislooked something somewhere, or falsely copied data from the gatekeeper log to windows notepad. Anyway, in front of me I have a trace from few days ago where the GCF contains 131.xxx.xxx.53 as rasChannel, and where as far as I remember the RRQ was sent to 131.xxx.xxx.69. Hence my astonishment. > > But if now I look into the GCF and the RRQ's IP destination address, they match whatever setup I use. > So whatever it was that caused the mismatch, now it is resolved. > > What I have concluded up till now is that setting Home= in the [Gatekeeper::Main] section does not force GCF messages to be sent out over that interface on a machine with multiple interfaces (instead it goes out over the machine's default interface), whereas the RCF is always sent from the IP specified by Home=. Should not also the GCF be sent out over this specified interface, such that always really only the one interface specified is used for all H.323 traffic? > > If no Home= is specified, although the gk tells the client to send RRQ to 131.xxx.xxx.69 (don't have a clue yet why it chooses so), the RCF is sent out via the machine's default interface 131.xxx.xxx.53. Is this a bug in the gatekeeper? The result: although the endpoint is now registered, OpenPhone does not know about it and reports a 'gatekeeper transport error'. > > What's left for questions from my initial mail is this: > Q4> can I specify anywhere the number of RRQs OpenPhone sends before deciding that registration did not succeed? > Q5> why does OpenPhone always send out two GRQ messages almost instantaneously when in automatic gatekeeper discovery mode (I did not set any reregistration entry)? > Q6> why does the gatekeeper send back 4 GCF messages if no home IP is specified and OpenPhone is in automatic gatekeeper discovery mode? > > Best regards, > Rene. > > > ------------------------------------------------------- > SF.Net is sponsored by: Speed Start Your Linux Apps Now. > Build and deploy apps & Web services for Linux with > a free DVD software kit from IBM. Click Now! > http://ads.osdn.com/?ad_id56&alloc_id438&op=click > _______________________________________________ > List: Openh323gk-users@lists.sourceforge.net > Archive: http://sourceforge.net/mailarchive/forum.php?forum_id49 > Homepage: http://www.gnugk.org/ ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ List: Openh323gk-users@lists.sourceforge.net Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549 Homepage: http://www.gnugk.org/
