Hi Pietro, if your trace files are too large, I would either write a (Perl ?) script to filter it down or search in it with an editor that can handle large files. If you still feel the sheer volume is too large, I would reduce the trace level. No need to take the performance hit of a level 5 trace if you don't use it. Since manually unregistering the MCU helps, I would switch down to trace level 3 for the next conference, so you can see all the RAS messages and you can check why the MCUs aren't able to re-register. Regards, Jan -- Jan Willamowius, Founder of the GNU Gatekeeper Project EMail : jan@xxxxxxxxxxxxxx Website: http://www.gnugk.org Support: http://www.willamowius.com/gnugk-support.html pierlu wrote: > Hi everybody. > > I have to explain a bit before asking my question. > > *This is the situation* > When I'm doing large videoconferences I have to cacade more tha one > MCU. These are the 5 MCUs I use: > > A: Radvision viaIp 400 20 ports > B: Radvision Scopia 24 ports > C: Radvision Scopia 24 ports > D: Radvision Scopia 12 ports > E: Tandberg 16 ports > > Services are set to 4CIF@386kbps max. MCUs B, C and D accepts also > CIF@384kbps and 4CIF@256kbps > > Cascading is done this way D -> B <- A -> C <- E ( -> indicates the > direction of the calls). So A is the master MCUs. Rooms are created by > dialing in. > > *This is the problem:* > It happens seldomly, with no apparent trigger event, that during the > conference MCUs B and C will completely fail. Clients connected to one of > this MCUs will experience the following situation: > > 1) first the audio will stop working. is lie conference is muted: this is > the sign for me that the MCU is about to go down. > 2) video will work nicely up until the call completely drops. > 3) clients trying to dial in the room again will get a denial and if me, > who's managing the MCUs, tries to recreate the room, gets the message > "Resource Unavalaible" from the MCUs boards, just like all of the usable > ports are used up, but actually no one is connected cos connections are > being refused (I cannot say whether they are refused by MCUs or by GnuGk > nor I can say whether this happens on LRQs or ARQs, my fault but the log is > too big to tell, read further). > > *This is what I don't understand * > Why do I have to unregister the MCU from the GnuGk with the unregisterip > command via telnet interface to regain a full functional MCU? Only by doing > so, media channels are newly avalaible and the MCUs will accept incoming > calls. Of course the first times it happened I hard reset the MCUs to > regain functionality, but in the end I discovered that its unregistration > from the gatekeeper that solves the problem (btw TTL is set on 60 seconds). > ** > *This is my question: do you think that, given the symptoms, fault is due > to the interaction of GnuGk and the two MCUs or that is to blame only > on the MCUs'? * > Consider that those two MCUs do not have any problems when handling > conferences on their own (without being cascaded, I mean). > > I tried to debug GnuGk log, but I'm using trace level 5 and when cascading > the log reports events for 80 clients and so it's pretty unreadable. > Moreover, I have no clue to what to look for. > > Next Monday I have another of this "big" videoconferences... do you have > any suggestion on what to look if this happens again? (as I told you, this > does not happen on a regular basis, it may not happen at all). > > Excuse the long explanation, and the lack of GnuGk configuration to go > along it, but I can provide them if you think you have a clue of what may > be going on. > > Thank you for your attention. > > Pietro Luigi Angelucci. ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________________ Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx Archive: http://sourceforge.net/mailarchive/forum.php?forum_name=openh323gk-users Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users Homepage: http://www.gnugk.org/
