Hi Michal,
thanks. This makes task for routing decision making systems much easier.
The problem with "h323-redirect-ip-address" is that radius (routing, billing) server needs to maintain its own table of currently active sessions per destination gateway in order to be able to issue proper h323-redirect-ip-address messages. However, in practice this is very difficult to achieve.
We tried different possibilities. When you base your session records on accounting stop and start messages it works fine, but problems start to appear when there are more than one call in setup stage, i.e. at the time when 2nd call for same destination arrives, 1st one may not yet be connected and there is no record in session table for this call. So routing system does not know that channel is actually taken and still sends call to that gw.
On the other hand, you may use access-request message to create session and accounting-stop to delete them. But in this case problems await in situation where call is not being authenticated at first attempt (due to network issues, slow radius response etc). So gnugk sends several attempts and there will be more than one session record per call.
Even not to mention situation when gk crashes (is killed, or network goes down) and accounting-stop messages do not arrive ever. Sessions ten can "hang" forever.
Basically this all is about unique identification of calls accross auth and acct messages - nothing new, same old radius problem ;)
Thanks for your excellent work anyway.
Zygmuntowicz Michal wrote:
Hi Aivis,
h323-redirect-number works more like [RasSrv::RewriteE164] and it takes step just before routing/outbound gw rewrite is made, so, with h323-redirect-number, gateway capacity limits apply as well as per-gw outbound rewrite rules. It overrides only inbound global/per-gw rewrite rules.
----- Original Message ----- From: "Aivis Olsteins" <aivis@xxxxxxxxxxxxxxxx>
Sent: Wednesday, November 10, 2004 5:14 PM
I suspected that.
So this also means that if "h323-redirect-number" is present in access-accept, the number will be sent out as specified by radius and any outgoing [RasSrv::GWRewriteE164] rules will be ignored, correct?
regards, Aivis
Zygmuntowicz Michal wrote:
I guess that if your system redirects the call to a specific IP
- it knows what it is doing. Therefore, nothing more (like matching
prefix, other gateways) is checked. The call is simply and unconditionally
redirected to this given IP.
----- Original Message ----- From: "Aivis Olsteins" <aivis@xxxxxxxxxxxxxxxx>
Sent: Wednesday, November 10, 2004 4:57 PM
Has anybody tried how radius based routing works in 2.2 when you define more than one GW per destiantion with limited capacity and different priorities?
Suppose GW1 has higher priority than GW2, but it is already filled up. Now radius server sends "h323-redirect-ip-address=<GW1>". Will call be routed to GW1 regardless of fact that it's capacity is full or will it respect Capacity setting and route to GW2?
Any thoughts?
Regards, Aivis
------------------------------------------------------- This SF.Net email is sponsored by: Sybase ASE Linux Express Edition - download now for FREE LinuxWorld Reader's Choice Award Winner for best database on Linux. http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
_______________________________________________________
List: Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549 Homepage: http://www.gnugk.org/
------------------------------------------------------- This SF.Net email is sponsored by: Sybase ASE Linux Express Edition - download now for FREE LinuxWorld Reader's Choice Award Winner for best database on Linux. http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
_______________________________________________________
List: Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549 Homepage: http://www.gnugk.org/
